“We can always tell” — No, you can’t.
--
A while ago, Guitarist, Youtuber, Call-In Show Host and generally cool girl Katy Montgomerie shared a clip found on TikTok of a cisgender woman being harassed in a pulbic restroom by another cisgender woman who apparently thought she was trans. The second woman tried to have the first thrown out of the toilets, although eventually backed down when the young woman’s boyfriend and the (male) security intervened.
The clip is on twitter here if you want to watch it for yourself. However, be warned; although it ends okay, it shows a fairly stressful confrontation in an enclosed space with raised voices and some transphobic language.
Anyway, the video got me thinking about how often anti-trans types claim that this simply doesn’t happen because they can “always tell” somebody’s “biological sex” just by looking. The claim is absurd on its face, and is more often than not just thrown as an insult to disparage or belittle trans and gnc people by making us feel uncomfortable or insecure about our appearances. So many gender non-conforming and trans people develop anxieties over how we look and how much unwanted attention we might be drawing to ourselves precisely because our safety in public is related to how well we blend in with the rest of society. So many of us have firsthand experience of what it’s like to be ‘clocked’ in public and then picked on, harassed or even attacked because of our gender identity and/or expression. Weaponising that to try and hurt the feelings of trans and gnc people through snide comments is a particularly vile bullying tactic in my opinion — simply trying to make people feel ashamed and insecure about themselves.
As I said above, it’s also clearly absurd: people are actually generally quite BAD at reading the sex/gender of strangers. Aside from the many, many examples of trans-hostile people getting it wrong (Mia Moore on twitter has dozens of examples you can see for yourself here), I have anecdotal evidence from throughout my life; I’ve been frequently mistaken for a girl: in person, from the front, from behind, from the side, over the phone… it’s simply not true that everybody has a perfectly infallible sex-radar. In fact people’s radars are so bad, that there are a lot more ‘false positives’ like the one in the video than there are ‘true positives’. Why? The reason is maths. (Haha! Got you! This is a maths article now!!)
Bayes’ theorem or bae’s theorem? 🤔
Let’s assume that people in general are really good at detecting trans people. Let’s say if you are trans then 9 out of 10 times, somebody else can tell that you’re trans with certainty. We call this, 90% sensitivity. And if you’re not trans, they can tell that with certainty 8 out of 10 times. We call this 80% specificity.
But both of these are the probability that a person says you’re trans, given that you are or are not trans. What about the other way round? We want to know probability that you really are trans, when a transphobe thinks you are? Well to calculate that, we use Bayes’ theorem:
where, P(A) is the probability of A and P(A | B) is the probability of A, given B, etc.
We want to calculate the probability that somebody is trans given that they’ve been ‘clocked’ (i.e. somebody thinks they’re trans), that is:
So let’s plug in some of the numbers we know. We know from recent census data in the UK that the probability of a random person being trans, P(trans), is around 0.5%, but for the sake of a thought experiment, let’s pretend that it’s actually 4x that number and say that trans people are 2% of the population, i.e P(really trans) = 0.02. We already assumed that people have a 90% sensitivity to detecting a trans person, so P(clocked | trans) = 0.9
Calculating P(clocked) is trickier…. what is the probability of getting ‘clocked’ in general? Well, it’s the probability of saying somebody is trans when they are, plus the probability of saying they are trans when they aren’t. Mathematically, that’s
(Note that we’ve used another slightly different form of Bayes theorem here which is that the probability of A is the probability of A given B, multiplied by the probability of B. i.e. P(A) = P(A|B)P(B) )
So, sticking that into our equation from above:
8%. There’s an 8% chance that when a transphobe says somebody is trans, they’re actually trans. That’s uhhh… not really very good huh? Maybe 90% sensitivity and 80% specificity are too low? Let’s be super generous and assume that people actually have 100% sensitivity and 90% specificity? Well, plug in the numbers yourself and you can see that bumps P(trans | clocked) up to a massive…. 17%. Hmm.
Even if sensitivity and specificity are both 98%, P(trans | clocked) = 0.5.
Even with 98% sensitivity and specificity, every time you guess that somebody is trans, you’d only be right 50% of the time… Why??
Well, it’s pretty simple really. It’s because there are way more cis people than trans people. In our example, 99.5% of people are cis. That would mean for every trans person, there are 199 cis people. And because of that, the number of misidentified cis people will pretty much always be larger than the number of correctly identified trans people.
In the 1st case (90% sensitivity, 80% specificity), if we have 1000 people, 980 of them are cis, but we get 196 false positives (0.2 * 980). 20 of them are trans and 18 of them give true positives (0.9 * 20). Which means that out of 214 people we guess are trans, only 18 of them actually are.
And remember, we assumed that 2% of people are trans. In reality it’s more like 0.5%. If you try doing these calculations again, you’ll see that makes things even more dire.
Transphobes need to be able to identify trans people and NOT misidentify cis people with simply unbelievable accuracy, otherwise the number of ‘false positives’ like in the video at the top of the thread will ALWAYS outnumber true positives.
Anti-trans activists love saying “we can always tell” and imagine thattheir ability to identify a trans person is 100% because they spend all day staring at pictures of people they already know are trans, but unless specificity is ALSO 100%, they’ll misidentify cis people far more often than correctly identifying trans people, simply because there are many many more cis people. As transphobes become more and more emboldened to harass people in real life, we’re going to see more and more cases of transphobes going aggro on cis people who they mistakenly think are trans because that’s just how numbers work.
I here want to re-emphasise though: harassing people over their gender, cis or trans, is f**king inappropriate. If you see somebody minding their own business in a toilet, don’t try and ‘transvestigate’ them, leave them alone!!
