A little less conversation, a little more close reading please: on D’Angelo and Marchiano’s response to Julia Serano on rapid-onset gender dysphoria
Lisa Littman’s study on rapid-onset gender dysphoria, initially published as a poster abstract in the Journal of Adolescent Health, recently came out in full-form in PLoS ONE. The article is making waves, because it purports to show the existence of a new developmental pathway for gender dysphoria, termed rapid-onset gender dysphoria, which would be distinct from traditional presentations of late-onset gender dysphoria and raise doubts as to the appropriateness of gender-affirmative care for a significant subset of transgender teenagers and young adults.
Shortly after the study came out, well-known trans author and biologist Julia Serano published a critique of the study, titled “Everything You Need to Know About Rapid Onset Gender Dysphoria,” which highlighted various flaws in the study by Lisa Littman which severely undermines its reliability and helpfulness in indicating the presence of a new developmental pathway.
A few days later, Roberto D’Angelo and Lisa Marchiano published a response to Julia Serano’s work on the website of the Pediatric and Adolescent Gender Dysphoria Working Group. The newly-formed working group is dedicated to the study of rapid-onset gender dysphoria and includes a range of controversial members such as Ray Blanchard, Susan Bradley, James Cantor, Debra Soh, Kenneth Zucker, James Caspian, and J. Michael Bailey.
Unfortunately, their response fails to adequately address many of the concerns raised by Julia Serano. I will comment on some of their claims before more broadly criticizing their portrayal of Littman’s work.
Sole reliance on parental self-report is scientifically unsound
Julia Serano pointed out that Lisa Littman’s study relied on parental self-reports in order to draw conclusions about children, making it inherently unreliable. According to D’Angelo and Marchiano, this does not invalidate the study’s findings insofar as obtaining information from parents is standard and essential practice in child and adolescent psychiatry.
Although their claim that it is standard practice is useful contextual information and adds some nuance to Serano’s portrayal, they fail to rebut her argument.
Firstly, just because something is standard practice doesn’t mean that it is unbiased. Sometimes, pragmatic and technical considerations constrain our ability to obtain unbiased observations. When this happens, we must maintain healthy scepticism about the data we obtain and adopt a critical outlook when interpreting this information, accounting for the ways in which it may be biased.
Secondly, sole reliance on parental reports in clinical contexts would be professionally negligent. However important parental reports are, clinical ethics require that information also be gathered directly at the source by interacting with the child. The problem with Lisa Littman’s study is not that she relied on parental self-report but that she gave undue weight to parental self-report by relying solely on it. A much more ethically and scientifically justifiable approach would have been to conduct concurrent studies of both parents and teenagers, as is regularly done in peer-reviewed publications, especially in psychology.
Thirdly, this critique fails to distinguish between reliance on parental self-report in clinical contexts and in research. Approaches which are justified in clinical contexts may differ from those which are justified in research contexts. The difference in power dynamics between clinical care and research has a significant impact on incentive to lie, for example. Parental reports in clinical care are often used to complement child report, often because clinicians suspect teenagers might lie to obtain care. In ethical research, responses from participation cannot impact access to care and therefore incentives to lie are absent. (Of course, it is also important to discuss how clinical practice and in particular gatekeeping in transgender care incentivises lying and compromises the therapeutic alliance. That is, however, tangential to the discussion of Littman’s study.)
Fourthly, this ignores the fact that Serano’s claim isn’t just that parental self-reports is unreliable, but that we have specific and pressing reasons to believe that the parents included in the study are biased in a specific direction. This isn’t an indictment of parental self-report in general, but rather a pointed and specific critique of parental self-report in the context of strong evidence selection bias.
Although parental report is standard practice in child and adolescent psychiatry, sole reliance on parental report in a study about alleged teenage psychopathology severely undermines scientific validity.
Sampling from anti-trans websites introduces significant bias
As Julia Serano pointed out, Lisa Littman’s study recruited on three websites which are well-known for their anti-trans focus. However, D’Angelo and Marchiano suggest that this does not undermine the study’s findings because it is standard practice to recruit from websites or groups where the target population is expected to be found.
It is indeed standard practice. However, as previously pointed out, standard practice doesn’t change the fact that bias may occur. Again, the critique isn’t that recruitment from a narrow range of websites is unscientific, but rather that the websites used give us strong reasons to believe that the study population was severely biased in a specific direction.
At this point, we must distinguish two critiques. The first one, which is the one D’Angelo and Marchiano try to address, is that the recruitment pool biased the sample. Although it is true that it biases the sample and gives us some pause when reading the data insofar as the parents may either lie or have their answers coloured by their ideological stances, the second critique is much more damning: even if we could be certain of the truth of the parents’ answers, the fact that they are parents we can expect to overwhelmingly hold anti-trans views is significant to interpretation.
The most impactful way in which the sample impacts interpretation is when interpreting the data on the degradation of mental health and parent-child relationships. In a sample where the parents are supportive of their children’s gender identities, we might suspect that something funky is going on: why would they be getting worse instead of better if transitioning was right for them? (Of course, social stigma and temporary fluctuations in gender dysphoria could play a role, but I am setting those aside for now.) But that’s not the sample we have and guess what? We already know that parental support for gender identity is a significant predictor of mental health. Children of parents who reject their gender identities — a rejection which is encouraged on websites like 4thWaveNow where Lisa Littman recruited participants — are much more likely to be suicidal. And of course, parents who reject their children’s gender identities are likely to see a degradation of parent-child relationships. If my parents rejected my gender identity, I would resent them and probably lash out at them.
Adolescent-onset gender dysphoria isn’t new
According to D’Angelo and Marchiano, “until very recently, adolescent onset gender dysphoria in [teenagers assigned female at birth] was almost unheard of.” This is somewhat but not entirely true. A more accurate claim would be to say that adolescent onset gender dysphoria, until recently, was very rarely observed in gender identity clinics.
It is true that people assigned female at birth who came to gender identity clinic as adolescents have seen a stark rise in recent years. There are multiplicity of reasons for this rise and for the shift in gender ratios. This change in clinical demographics was implicitly predicted by practitioners decades ago, and should not come as a surprise given the stark increase in trans visibility in the general population in the last decade.
Because this text is already long enough as it is, I won’t go into details of why clinical populations are shifting. Since the Littman study isn’t a clinical sample, it seems more appropriate to compare it to non-clinical data on trans people.
Adolescent-onset gender dysphoria isn’t all that new. As many as 40% of trans adults began feeling like they may not be cisgender at or after 11 years old, and 17% report beginning feeling that way between 16 and 25 years old. Barely 5%, however, began telling others they were transgender prior to 11 years old.
However rare teenagers assigned female at birth have been in gender identity clinics prior to the last decade, they are far from unheard of in the broader trans population.
PLoS ONE is a respectable journal, but…
Whether PLoS ONE is a respectable journal depends on what you consider respectable. Publications in it are frequently cited, and a number of excellent studies can be found in it. It does, however, have an unusually high rate of publication: it publishes over 50% of submissions, compared to roughly 11% for the Journal of Adolescent Health, which would have been a respectable and well-suited journal for the study topic.
Independently of respectability, the fact that it was published in PLoS ONE is significant given the crux of the criticism. PLoS ONE does not, for the most part, vet study interpretations. It primarily vets the quality of the statistical analysis. If the data is well translated into statistics, PLoS ONE is happy.
However, critiques of the study aren’t that it misrepresents the numbers, but that it misinterprets them. Data framing and data interpretation is the bulk of the work in publications. They’re what is relied upon by policymakers and what is most often taken as “the finding” of the study.
Julia Serano’s critique of the study is pointed. It is specific. The study’s observations is better explained by what we already know about trans people and specifically trans teenagers than by positing a new developmental pathway for gender dysphoria. The burden of proof is on Lisa Littman to show why non-pathological interpretations of the data are less adequate than interpreting it as evidence of rapid-onset gender dysphoria. Reviewers from journals other than PLoS ONE would have likely challenged Littman to explain why she interpreted the data in the way she did when other, more plausible explanations were available.
That some studies published in PLoS ONE are fantastic doesn’t address the fact that an integral function of scientific journals is to hold authors accountable for their interpretations and that PLoS ONE failed to fulfil this function.
The study’s data doesn’t suggest social contagion
A core argument made by D’Angelo and Marchiano is that although Serano is right in highlighting that Littman’s study doesn’t prove that social contagion causes rapid-onset gender dysphoria, the study certainly strongly suggests it.
Their response ignores how Serano’s arguments contradicts the claim that the data even suggests social contagion as a cause. Littman’s data is expected given what we know about trans teenagers. Language “copied verbatim from online sources” is hardly new. It is a well-recorded fact of trans history that trans folk have relied on mutual support in coming out to their closed ones and in accessing trans health care. Already in the 70s, trans women were using Harry Benjamin’s book to convince psychiatrists to grant them transition-related care. And in what can be best described as the least surprising plot twist of all time, teenagers who question their gender identity tend to navigate toward websites and friend groups which discuss gender issues.
To put my response in more scientific terms, Littman must rebut the null hypothesis if she is to claim support for the presence of social contagion. Since the null hypothesis — that these are unexceptional adolescents with late-onset gender dysphoria who weren’t victims of social contagion — is a hundred percent compatible with the study’s observations, the study fails to provide any suggestion whatsoever of the presence of social contagion. The most we can say in favour of the thesis of social contagion is that it was neither supported nor disproved by the study: it is agnostic in that regards. And so we fall back on the null hypothesis.
Co-morbid mental health issues among sampled youth weren’t abnormal
Contrary to D’Angelo and Marchiano’s suggestion that the youth represented in the study showed abnormal levels of mental illness, the level of mental illness was in line with the prevalence of mental illness among trans people.
Although “Littman found that 62.5% of young people were diagnosed with one or more psychiatric disorder prior to announcing that they were trans,” it is well worth noting that the overwhelming majority of diagnoses were for anxiety or depression, who mental health problems which are common among trans people are strongly correlated to social stigma, untreated gender dysphoria, and lack of perceived parental support for gender. And although more severe mental health concerns are present in a non-trivial number of teenagers, the rate is in line with the rate of co-morbidity we would expect in a sample of youth with gender dysphoria. Since parents of teenagers with severe mental illness seem more likely to be concerned about the appropriateness of transition, there are also some self-selection concerns with regards to those statistics though I do not believe the impact to be statistically high.
As for the high rate of traumatic experiences, the inclusion of mundane events such as break-ups in the definition of trauma makes the reported percentage wholly unreliable. All-in-all, the rate of mental health issues among the youth represented in the study is expected.
Many of D’Angelo and Marchiano’s response to Serano’s methodological critiques share the same structure. “X is sometimes or often appropriate, and therefore it was not a significant concern in the Littman study.” However, this relies on a fundamental misreading of Serano’s critique: she isn’t saying that X is always inappropriate, but that it is inappropriate in this case for specific, detailed reasons.
Although the two authors agree that it “would be unreasonable to conclude that the study provides irrefutable proof that ROGD is a distinct entity,” they nevertheless suggest that it provides support for the existence of ROGD. Therein lies the problem with their critique: the study, because it fails to rebut the null hypothesis that the studied children are a relatively unremarkable group of teenagers with late-onset gender dysphoria whose parents hang out on anti-trans websites, fails to provide adequate support for the existence of ROGD. It doesn’t prove it exists, and it doesn’t support it either.