On small differences, academics, and biological sex
by Danielle Navarro, 12 Jul 2020
Epilogue and prologue:
After I posted this essay, one of the authors of the article I critique here – Bill von Hippel – contacted me via email to discuss some of the issues I raised. We had a polite exchange over a few weeks, covering a range of topics surrounding transgender inclusion. I won’t rehash the conversation we had, but because my essay below is very critical of his opinion piece, I thought it is only fair that I include the happy ending. After a long discussion, this is what Bill ended up saying to me:
I understand this must get tedious for you to have to re-explain to every neophyte who comes along … I find your arguments [in our email exchange as well as in the post appended below] very compelling, and I’m now convinced that I was in error to raise the issue of transgendered people at all in our article. I apologise.
I want to mention this at the beginning because it is unusual in my experience that someone would make an acknowledgment like this, and I respect and appreciate the fact that Bill did so. I wish it were more common.
There is a strange kind of reasoning failure that sometimes afflicts academics when the subject of transgender people arises. It manifests itself – in newspaper articles, in blogs, on social media and in assorted other places – as a regrettable excess of hyperbole. Academics as a rule are not incautious people. Yet on the subject of transgender people, esteemed professors who might under normal circumstances speak in a careful and measured fashion suddenly become prone to wild exaggerations and reckless generalisations. I don’t know why this happens, but I have some hypotheses.
A wonderfully perfect example of this crossed my path yesterday, quite unbidden. An email appeared in my work inbox, forwarding for my attention an article recently published in The Australian newspaper. With some trepidation – knowing only too well that articles published in that outlet are rarely sympathetic to transgender people – I read the piece, and went through quite the range of emotions. At first I was angry, then then miserable, and then frustrated. I was angry because of what the article has to say, miserable because it was sent to me out of the blue, and then frustrated because I couldn’t work out how to explain my concerns to my academic colleagues in terms they might understand. This post is my attempt to do so.
Although the article is published in a national newspaper, the authors are academics and they self-identify as Madeleine Beekman, a “professor of evolutionary biology at the University of Sydney”, and Bill von Hippel, a “professor of evolutionary psychology at the University of Queensland”. The article begins with this passage:
The Australian Academy of Sciences recently changed its definition of a woman. According to the new definition, anyone who identifies as a woman is a woman, regardless of their biological sex.
This does sound quite dramatic, but as usual the reality is rather more bland. What this passage refers to is the fact that the Australian Academy of Sciences (AAS) recently released their Women in Stem Decadal Plan, a document of 66 pages that contains a working definition of the term “woman” in a glossary. On page 58, at the very end of Appendix 5, the document notes that they include transgender women within the scope of the term “women” for the purposes of the Decadal Plan. It’s not a particularly radical document, and it doesn’t actually address the concerns that transgender women might have. The word “transgender” only appears in one other place, as part of a paragraph noting that LGBTIQ women face particular difficulties in STEM. Reading the document as a whole, one does not get the impression that the AAS is taking a particularly strong stand on the metaphysics of gender. Rather they are acknowledging that transgender women – such as myself – deserve some consideration. Not quite the ideological extremism one might have been expecting based on Beekman and von Hippel’s description.
In any case, the authors continue:
This definition has the clear advantage that people who don’t identify with their biological sex will now be recognised as their preferred gender, an obvious social justice issue.
“Oh how nice”, I think to myself. Then I read the next sentence and my heart sinks.
However, with this new definition of woman, the academy is tacitly stating that biological sex is of no significance.
This seems rather at odds with the substance of the Decadal Plan, but alas the reader is given no opportunity to discover this for themselves, and in any case the article immediately pivots away from transgender women. The vast majority of the piece has nothing at all to do with transgender people. Instead it focuses on whether there are meaningful biological sex differences between men and women that might explain the observed gender disparities in STEM disciplines.
As a scientist as well as a transgender woman, I confess that I am puzzled and more than a little annoyed by what I have just read. Though the AAS document does in fact discuss many issues that pertain directly to biological sex (maternity leave, for example, appears to have some conceptual relationship to biology at least in my shallow understanding of the subject) the Beekman and von Hippel piece quite boldly asserts the contrary:
However, with this new definition of woman, the academy is tacitly stating that biological sex is of no significance.
This is patently absurd, and if I hadn’t seen this kind of ridiculous exaggeration from academics so very many times before when they are faced with the revelation that transgender people exist I would be surprised that intelligent people would make this error. Indeed, when I point out that this is an utterly unreasonable thing to say, I am sometimes met with blank surprise and defensiveness. The claim that this article is asserting is pure nonsense, but many people seem unable to see why.
So I suppose I had best explain why it is a nonsense. I wish I did not have to. I am very, very tired of this silliness.
Let us start at the beginning. As I mentioned above, the AAS document is pretty anodyne and barely discusses transgender people at all. However, what it does do is include a glossary entry for “woman” that is inclusive of transgender and nonbinary people:
Anyone who identifies as a woman, including cisgender (personal gender identity corresponds with sex assigned at birth), transgender (personal gender identity does not correspond with sex assigned at birth), non-binary and intersex persons who identify as a woman (or girl)
It is this passage that has so inflamed passions, and it is on the basis of this passage that Beekman and von Hippel assert that the AAS is tacitly stating that biological sex is of no significance. This seems like a rather silly overgeneralisation to me.
To see this, suppose for the moment that AAS had not adopted a trans-inclusive definition. Suppose they had decided to exclude transgender women, and treat “woman” as a purely biological category. They might, for example, have decided to define “woman” solely by reference to chromosomes. They might have decided to assert that people with XX chromosomes are women for administrative purposes and people with XY chromosomes are deemed men. It’s a view I see expressed a lot on twitter, but any biologist would be quite quick to note that classification based on this simple rule would cause some difficulties. Biology – like any complex system – is somewhat messy. Some people are intersex. The crude chromosomal classifier would produce a lot of counterintuitive and undesirable decisions if applied to intersex people in a thoughtless way. As a consequence, some care is required, and indeed the AAS definition does take due care not to be unkind to intersex people.
That being said, it should be noted that there really aren’t that many intersex people. They make up a smallish proportion of the population (only about 1-2%) so even the crude “XX = woman, XY = man” rule that I see so often on twitter would actually classify people correctly maybe 98% of the time (that is, assuming that we are willing to disregard transgender people). If we viewed this as an abstraction and did not care about the lives of the affected people, we might be rather pleased with this: 98% classification accuracy is pretty good in most situations.
However, because we are not jerks we recognise that a little care is needed. The rights of of intersex persons are important, and so rather than rigidly and unwisely insisting on an overly-simple rule we aim to be flexible. In most circumstances our approach is in accord with the AAS definition: we respect the autonomy of intersex people and classify them as members of the gender category that they feel is most appropriate. Moreover, the fact that we do so is genuinely unremarkable. It is rare – unheard of in my experience at least – for academics to become aggrieved over such accommodations. There are no articles written in national newspapers taking up arms against the scientific and metaphysical horror that would be unleashed if we started being kind to intersex people.
Why, that would be absurd.
The existence and recognition of a tiny minority does not suddenly render the ordinary biological categories “of no significance”, now does it? Only a shallow and overly rigid thinker would make such a silly error.
Dear reader, you are not foolish and I know that you can see where I’m going with this, but please bear with me as I go through the motions, tiresome though they may be.
Suppose we now consider what happens when we adopt a trans-inclusive definition of gender categories, as the AAS has done. Like intersex people, transgender people are a very small minority. Estimates of our prevalence vary, but for the sake of this essay I’ll assume that 1% is a reasonable number. It’s not large. Even taking into account the recent shifts in attitudes towards transgender people, very few people who were designated male at birth will choose to live as a woman. Very few people designated female at birth choose to live as a man. Some people choose to reject both categories and identify as nonbinary, but again their numbers are few.
What does this entail for our simplistic chromosome-based classifier? If we choose to respect transgender people in our definition of gender categories, it’s going to perform slightly worse. It will now misclassify some transgender people in addition to the intersex people it misclassified in the previous scenario. The overall classification performance might shrink from 98% to 97% accuracy. Not a very large difference at all, wouldn’t you agree?
Yet this leads me to considerable confusion, or at least would do if I chose to believe Beekman and von Hippel’s interpretation of the AAS definition. The first (traditional) definition of gender implies about 98% prediction accuracy when inferring gender from sex (or vice versa). The second (trans inclusive) definition implies about 97% accuracy. How can any rational person believe that the first of these two scenarios implies that biological sex has an important relationship to gender, while the second one is a tacit claim that biological sex is of no significance at all?
Why, that would be absurd.
When framed in terms of the empirical base rates of intersex and transgender people, the scientific illiteracy of Beekman and von Hippel’s claim is painfully clear. For those among us who choose to live in the real world and adopt a “cold hard facts” perspective it makes no sense whatsoever to assert one definition ignores the reality of biological sex and the other does not. When the practical differences between the two approaches amount to a mere 1% difference in classification accuracy, it simply cannot be true.
Given the strangeness of this error it becomes pertinent to ask: why do smart people make such foolish mistakes? I have several hypotheses, none of which I’ve ever been able to confirm or falsify very well.
The first hypothesis might be called the blind spot theory. According to this hypothesis, the error arises due to a kind of thoughtlessness or cognitive myopia. Because transgender people are so rare and our experiences are so very different to those of cisgender people, it would hardly be surprising if cisgender people were not skilled at thinking through the entailments of their claims, when those claims relate to our lives and not theirs. Or, to put it crudely, the blind spot hypothesis is that cisgender people simply do not think about us at all.
How would the “blind spot” hypothesis work in this scenario? What I might imagine happening is that a cisgender person might look at the AAS definition – or worse, a newspaper article or press release referring to it in too-simple a fahsion – and think something like this: “Well anyone might self-identify as a woman with no consequences, right? And if that were to happen there would be no relationship between sex and gender at all!” From there, one might quickly spiral into hyperbole and moral panic.
The silliness of this reasoning is painfully obvious to every transgender person: there are consequences to being trans, and those consequences are very significant indeed. Even without looking at the (dire) statistics on transgender lives, it should be obvious to anyone that there are massive disadvantages to being trans. Compared to most trans women I am very lucky, but even for me it is grim: I regularly receive abusive comments on social media, I have been threatened with violence for being trans, and I live in constant fear that my basic rights will be taken away from me for no reason other than naked partisan gain by cynical politicians.
That is to say that identifying as a woman is not something one does lightly, and it does not come without risks. It doesn’t mean that on a whim I decided to throw on a dress one day (and in point of fact I don’t wear dresses very often). In my case, over the last four years I have
- been diagnosed with gender dysphoria
- taken hormone replacement therapy and, as a consequence
- grown breasts
- had my facial hair removed in an extremely painful way
- changed my name legally
- changed many many id documents and been repeatedly humiliated and verbally abused on occasions when doing so
- et cetera et cetera.
You get the idea. Of course, not every transgender person will have the same list. We vary quite a lot, and “transgender” is itself a very broad umbrella term that encompasses many different kinds of people. Nevertheless, I think it is fair to say that for (almost) every transgender person there are substantial consequences, and in most cases the practical result is that bad things happen to us. In contrast to the hypothetical worlds imagined by many cisgender people, the real world imposes a lot of impediments for transgender people. It is perhaps no surprise then that there are very few people who even want to transition, and fewer still who take any substantial steps towards doing so.
In other words, when we examine the practical consequences of the AAS definition in the real world, the idea that “anyone could identify as anything” doesn’t really hold up to any kind of empirical test. Yet, the only way it is possible for the AAS definition of gender to become “untethered” from biological sex is to imagine exactly such a world. There is little prospect of such a thing occurring in any of our lifetimes. The idea is terribly silly.
Unfortunately, in order to see that this idea is silly, you actually have to think it through. For transgender people this is trivially easy: we can look at our lives and it just “pops out” at us. But for cisgender people it’s not so obvious. They seem to have this blind spot that prevents them from seeing how absurd a claim this is.
So that’s the first hypothesis.
The second hypothesis I have that might explain the phenomenon is one that every academic will find very familiar: pedantry. Academics are notorious sticklers for “the principle of the thing”, are prone to dig in when we believe ourselves to be “technically correct” and so on. The pedantry account is subtly different to the blind spot account. According to the blind spot idea, cisgender people simply fail to think things through. According to the pedantry account, in contrast, a cisgender person might succeed in thinking through the practical ramifications, but simply decide that these practical considerations are irrelevant to the matter at hand.
Here’s how that account would work in this scenario. As I’ve pointed out, a practical comparison between the AAS definition and a traditional one – when viewed in light of the cold facts of the world we live in – reveals there is little difference between them. They classify people into gender categories in an almost identical fashion, with a mere 1% (give or take) of people making up the discrepancy. A true pragmatist would conclude that the difference doesn’t matter very much except to that 1% of people, and not make a fuss about treating the 1% with kindness.
But to the pedant, it is a different story. To the pedant this tiny difference in classification accuracy is suddenly important! It is the principle of the thing that matters! According to (their choice of) principles there is a philosophical difference between the two rules. And this philosophical difference – quite unlike the dignity of transgender people, evidently – must be respected. For you see, a traditional definition constructs a link between sex and gender in an explicit fashion. The fact that a relationship between the two things is literally written into the definition. Under the AAS definition, however, the connection between sex and gender is merely implicit. The strong association between sex and gender is of course maintained in the AAS definition, but it is now an empirical fact (i.e., because transgender people are rare) rather than an axiomatic truth.
To the pedant, apparently, such things matter. Yes, the pedant might concede, in our world as it exists these two rules are almost indistinguishable. But that is not the point! If the base rates were very different then the two rules would behave very differently. If we lived in a world where 40% of people were transgender then there would be quite marked practical differences. Ergo, we must treat this as a matter of importance!
To my mind, as a transgender woman who has become incredibly tired of the whole thing, this is all quite funny. In my experience the kind of person who makes such an argument also tends to fancy themselves a hard nosed scientist. “Facts don’t care about your feelings!” they declare, when dismissing the concerns of transgender people. It is puzzling however, because base rates are indeed empirical facts, and they don’t care about your feelings either. You cannot simultaneously claim to be a rational empiricist while indulging in wild hypotheticals about counterfactual worlds with 40% transgender populations. The (approximate) base rate of 1% is a fact, and it is one you are not entitled to ignore.
It seems to me that if it is pedantry that underpins such behaviour, it is pedantry of a particularly unfortunate – and hypocritical – kind.
My third hypothesis for this phenomenon might be called the clickbait hypothesis. According to this view, what is happening is that the authors don’t really care one way or the other about transgender inclusion. They actually want to talk about something else, in this case sex differences as they pertain to the (mostly cisgender) human population. However, this subject is … well, it’s kind of boring. It doesn’t exactly get people excited. So, because editors like to sell papers and academics love an eager audience, what the article needs is a cute or dramatic anecdote to “suck the reader in”. In other words, perhaps those paragraphs about transgender people weren’t meant to be taken seriously, they were just meant as “rhetorical flourish”.
What puzzles me about this hypothesis is that when I suggest it as a possible explanation, cisgender people jump at it. “Yes, yes!” they say “don’t take it seriously”. They breathe a sigh of relief, thankful that there is no cause for alarm.
They say this and I think… “But that’s actually worse. You do know that this makes it worse, right?” Under the blind spot hypothesis, the behaviour is accidental. We all have blind spots, and it is not hard feel sympathetic. Under the pedantry hypothesis, the behaviour is intentional but still slightly excusable. After all, we all have our moments of pedantry. But the clickbait hypothesis postulates that the writer knows perfectly well that their claims are unfair and somewhat cruel to transgender people, but they have deliberately concluded that it is justifiable for them to be unkind to us, if the result is that they get a few more clicks on their article. That is immoral and selfish.
There is one last hypothesis that I have. One I hope is not true. A hypothesis of last resort if you will. Sometimes people are hurtful towards transgender folks not out of blindness, not out of misguided pedantry, not even out of selfishness.
Sometimes they do it out of malice.
[Author note, 2020-07-13: This essay has been edited for clarity]