Evidence gaps: A reply to Buxton Part 2
In my last post I argued that those scholars and activists like Rebecca Buxton who go out of their way to emphasize the oft-cited figure that ‘only’ 1% of people who decide to undergo a sex/gender transition end up detransitioning, fail ethically. In this post I’ll argue that they fail epistemically too. This is because the evidence Buxton pointed to is not good evidence for the claim she is making.
As a reminder, this is what Buxton tweeted:
“I shared eight peer-reviewed studies on detransition/regret rates with someone in the comments on Rowling's post. They said it was all based on data faked by the scientists. I have no words.”
“He kept saying that I was coming up with “research” to support my own view. PEER-REVIEWED LONGITUDINAL STUDIES.”
The detransition rates in these studies vary from >1% – a little over 2%. I’m going to use a rate of ~1% in this blog post for simplicity sake.
I will only be able to provide a sketch here of the sorts of evidential issues I think Buxton and others are neglecting, rather than a full analysis of the evidence base using the tools of philosophy of science and science and technology studies. But before we we can start to assess, even at a high level, whether or not the evidence Buxton points to supports the claim she's making, we need to understand what, precisely, the hypothesis or claim she's actually making is.
Is she claiming that ~1% of people who, at some time of their lives transitioned in some way, either medically or socially, ended up detransitioning? Or is the claim that ~1% of people who have at some stage taken puberty blockers and/or cross-sex hormones end up detransitioning? Or is it that ~1% of people who had surgical procedures to alter their sex-based features ended up detransitioning? Or that ~1% of people who have had particular surgical procedures, like gonadodectomies, end up detransitioning?
These are all very different claims, which require different evidence to support them. The studies Buxton links to explore a range of different types of (de)transition, so the extent to which they even play a corroborating role for each other is unclear.
But notice that whatever the claim, these are all backward-looking time-bound claims: they refer to study populations of people who have already transitioned and detransitioned by a particular point in time. It's much easier to say that a claim is or is not accurate when we're looking only at a particular population of people, at a particular point in time, and that time is in the past.
What's much more difficult, of course, is to make an accurate prediction about the future. And it's a (forward-looking) prediction, not a backward-looking claim, that I think Buxton and others are actually performing when they emphasize the 1% claim. They're essentially accepting the following hypothesis:
1% of the people who transition today will end up detransitioning at a future point in time
But does the evidence marshaled by Buxton provide her with a good reason to accept that hypothesis?
Before we look at the evidence Buxton did point to, i'm going to make the hypothesis more specific, by disambiguating 'transition' and indexing events to a particular time period and place. This is to make it claim context-specific, which is important to evaluating evidence.
Let's only consider those who commence a medical transition, i.e. those persons who engage one or more of the following pharmacological or surgical interventions as part of a sex/gender transition: puberty blockers; hormone replacement therapy; one or more sex-reassignment surgeries. Note that it would likely be even easier to reach '1%' if social (de)transitions were counted. And let's index the hypothesis to the United Kingdom, and to the year 2020.
The hypothesis then becomes:
H: ~1% of the group of people who commence a medical transition in the year 2020 in the United Kingdom will end up detransitioning at some point during their lives
Now we can ask: to what extent does the evidence Buxton points to support her acceptance this hypothesis? I will suggest: nowhere near as strongly as she seems to believe it does [1]. In short, this is because important contextual features have changed significantly in recent years, such that there are good reasons to question the relevance of studies based on past time periods and cohorts to questions related to sex-gender transitions of cohorts in 2020.
———
One of the studies Buxton provides a link to is a 50 year Swedish longitudal study that found a 2.2% regret rate for both male and female persons who undergo a sex/gender related transition. This is an impressive data collection effort to be sure. But we have some good reasons to think that the period over which this study was conducted (1960-2010) may only be weakly relevant to more recent time periods, and to attempting to understand the range of plausible detransition rates for the cohort who commenced a medical transition in 2020.
Firstly, in many countries in Western societies, including the UK, it's only been since ~2010 that we've observed the large uptick in the number of people attending gender clinics. For instance, in 2009/2010, the Gender Identity Development Service (GIDS) in London only had 2 referrals per week, whereas in 2017/18 it was 25 times that number, up to 50 referrals per week. The most marked increase occurred in 2015-16, when patient numbers doubled compared to the previous year. Especially given that similar trends have been observed in other Western societies, it’s likely that the context prior to 2010 was very different to the post-2010 context. It’s far from clear, then, that the extrapolation of the results of studies pre-2010 to the post-2010 period is warranted.
An important contextual feature that differs between the period prior to 2010 and post 2010 that counts against this particular study's relevance to the hypothesis of interest is the make up of the study population compared to that of recent populations attending gender clinics. Recent observed trends highlight a marked change in the numbers and proportions of those 'assigned (observed) male at birth' (A/OMAB) versus those 'assigned (observed) female at birth' (A/OFAB) attending gender clinics compared to previous eras. In the Swedish study Buxton cited discussed above, the study population consisted of 478 A/OMABs and 289 A/OFABs. Similarly, another study Buxton pointed to as evidence involved a Netherlands-based study population over the period 1972-2015 consisting of 4,432 A/OMABs and 2,361 A/OFABs. However, between 2008-2018 there was a 1,500% rise in gender dysphoria diagnoses among 13- to 17-year-olds born as girls (A/OFABs) in Sweden. The number of those born as girls being referred for treatment rose 4000% in England and Wales over a similar period. A/OFABs now significantly outnumber A/OMABs in many gender clinics in the West. Thus it's unlikely that results derived from study populations predominantly composed of A/OMABs will tell us anything about likely rates of detransition in contemporary gender clinic populations, particularly those that deal exclusively with children and adolescents.
Given that demand is currently vastly out-stripping capacity at many of these gender clinics, it's reasonable to think that under these sorts of circumstances, the rate of false-positives may increase. People with less clinical experience and expertise may be being brought on to cope with the demand and staff are pressured to get people through the system to make way for more patients. Staff at GIDS in London complained about this, among other issues, in this recent BBC investigation.
The studies Buxton links to also deal with a variety of different types of transitions – sometimes only looking at (and hence making claims about) surgical interventions, and not other types of transition. If you’re interested in detransition rates amongst those who commence cross-sex hormone use—especially given recent trends in ‘micro dosing’—then it’s not clear that the detransition rates of those who undergo surgeries is relevant to those who only take hormones, and especially those who ‘microdose’.
What’s the upshot then in terms of what the detransition rate is likely to be for the cohort of British persons who commence a medical transition in a certain direction in 2020? I don’t actually think we have good evidence, at this point in time, to be making any strong claims about that. I think we would we need better analysis of the existing data’s relevance to current populations alongside further research to fill in our gaps in understanding, to be in a position to make more confident claims in any direction. I think it’s possible that it could remain the same or be lower; I also think it’s possible that it could be many times higher.
In the mean time, scholars and activists should scruntinize the science and not just ”read the science” before they cite it. In doing so, they should reflect on the extent to which they can credibly and ethically use this research to do the (political) work they want to use it to do.
——- If you’ve never heard detransitioners talk about their experiences before, here are a couple of incredibly eloquent women talking about their experiences:
Mackenzie (USA-based) here Watson (UK-based) here
——-
Endnotes:
[1] There are further technicalities here, because a scientist is unlikely to accept a forward-looking prediction outright and make (forward-looking) assertions about it. What she is more likely to do, is assign the hypothesis a probability, based on the evidence available to her, and her reasoning about that evidence (e.g accounting for known unknowns, etc.). So, to be charitable to Buxton, let’s say what she accepts is that it’s highly likely (e.g. >90% probability) this hypothesis is accurate (to e.g. +/– 1%).
This act itself, many philosophers would argue, is equivalent to accepting a hypothesis about a probability or probability distribution, and that that decision will itself inevitably involve the use of non-epistemic value judgments (i.e. involve the use of moral, social and political values in reasoning about the hypothesis). This is because, the theory goes, the decision to accept or not accept a hypothesis about a probability will itself be underdetermined by the evidence. So, these philosophers may say Buxton’s political values (the weight she places on trans persons being able to transition as early as they would like) justify her accepting and asserting this hypothesis.
There’s a more complicated analysis to be made here about what someone is or is not ethically permitted to assert in light of the evidence base and the politics at play. For those who think that Buxton is justified in shifting her epistemic standards (lowering them) because of her political values in this way, please see my first post.