Sometimes it seems like we, as a people, are incapable of discussing anything rationally, and it always seems to get worse when the discussion involves those who are in some way different from the majority. Heated accusations are frequently thrown around whether or not they have any basis in fact, or even make sense at all. And even when claims come across as calmer and more reasonable, they often end up being widely shared with little thought given to their accuracy or relevance.
One of the more popular of such claims is that most supposedly transgender children—80% according to 1995 findings by openly trans-antagonistic Zucker and Bradley, or 84% according to an oft-cited 2011 Steensma study, or even "as many as 95%" in the words of the president of a misleadingly-named fringe group—will naturally desist, that is to say, that they will end up being perfectly fine with their assigned-at-birth sex once they grow up. They're just going through a phase, or aren't mature enough yet to understand such things, or are even being pressured into assuming a nonstandard identity, the argument goes, so there's no reason to humor them. Those using this argument may additionally be trying to imply that even the remaining 20% (or 16%, or 5%) aren't really transgender, if such a thing even exists, but just want attention or are suffering from some kind of delusion. And even those among the desistance-allegers who concede that transgender people may be real would seem to be pushing the viewpoint that there's no way of telling who's going to grow up to be transgender and who isn't, so it's counterproductive or at best pointless to try treating transgender children accordingly.
There are several fairly obvious problems with this line of argument, even before getting into the many reasons the 80% figure (or whatever other number) is meaningless. If children who think they're transgender are just going through a phase, then they're just going to grow out of it regardless, so why not humor them? If they just don't understand, why not let them find understanding, and perhaps learn something valuable in the process, through harmless experimenting? If they just want attention, why aren't they instead using any of the innumerable easier, safer, and far less inconvenient ways of getting it? If they're facing pressure that leads them to claim that they're transgender, what about the far more prevalent pressure to not be transgender, often from the same clinicians who ran these studies? Speaking of which, considering the prevalence and intensity of said pressure, can we truly say that any observed desistance occurs naturally, or even be sure that the presumed desisters haven't merely gone into the proverbial closet?
Additionally, even if 80% do desist, this also means that 20% don't. Should we just ignore them? That seems callous and shortsighted. Furthermore, while a rate of 20% may seem low, it's significantly higher than the less than 1% estimate for trans people in the general population. So is even the apparently-made-up 5% figure. That suggests a meaningful correlation. If such a correlation exists, then apparent transgender tendencies in children cannot simply be dismissed as meaningless, which doesn't seem at all like something that the people making this claim would want to concede.
Worse, people sometimes treat studies that started with young children as though their results applied equally when considering teenagers already going through puberty. It should be obvious why this makes no sense, but leaving out important details can go a long way, especially when it results in something that appears to support already-decided opinions.
And while the argument is already problematic enough in and of itself, its central claim that most transgender children desist doesn't stand up to scrutiny, either.
To begin with, the studies cited to argue desistance suffered from very small sample sizes that limit any ability to draw any meaningful conclusions from them at all. The Steensma study, for example, included only 53 total children, and that was one of the larger ones. Given the dubious selection criteria (as explained below), it's unlikely that even ten of these children were properly gender dysphoric (which is still more than the total number of subjects in some studies that are cited). While that may be enough to suggest possible patterns that may call for further study, it's not nearly enough to generalize anything. Furthermore, the fact that the studies typically selected subjects from the children already being brought to their clinics may also have biased the study populations, and therefore the results. Parents, one would expect, would be more prone to going out of their way to bring children with non-conforming behaviors to a clinic if they viewed these behaviors as something that needed to be fixed, potentially by coercion or force if need be, than if they regarded these as harmless quirks. This would presumably have been particularly true in the case of clinics like Zucker's—before finally coming under scrutiny and eventually being shut down, it had engaged in something that at the very least strongly resembled the discredited practice of conversion therapy, which attempts to force people into normative behavior by any means necessary.
However, even if these had been large, highly representative samples, and even if we could be certain that every single child in every one of these studies was actually gender dysphoric, the 80% figure would still be misleading at best. That's because these results assume desistance for all of the subjects who didn't come back for follow-up years later, based on the reasoning that if they didn't return, they must not be having further issues. It's unthinkable, apparently, that they might have moved, or found a more supportive or just different clinic, or had second thoughts about participating in the study, or been dealing with related or unrelated issues that took priority, or died to assault or suicide or random chance, or that they did transition and found that it resolved their dysphoria (which I'm fairly certain is basically the point of transitioning). Assuming that no-shows must have desisted is a bit like assuming that anyone who stops going to one specific mechanic for service must have desisted from owning a car. This is compounded by just how large the affected portions of the study populations often were. In the Steensma study, for example, 24 of the 53 subjects never returned. That's nearly half! It would make just as much sense to assume that all of these subjects ended up transitioning and never looking back, which, just by itself, would instantly drop that supposed 84% desistance figure to a mere 40%. It would be more honest to simply exclude them from the results, though that, of course, would make the already problematically small sample sizes even smaller.
The returning subjects who were seemingly not dysphoric don't necessarily mean much, either. After all, they were typically checked on only once, most often as young adults, and not repeatedly evaluated over time, much less followed on an ongoing basis for the rest of their lives. It's easy to find examples of people who, whether out of fear of rejection, financial concerns, lack of information, or any of a number of other reasons, don't transition until their thirties, forties, fifties, or even later in life. It doesn't seem that any of these studies attempted to explore what caused the apparent change of mind in the subjects in this group, either. They might have simply grown out of a phase, yes. But, assuming that they were dysphoric to begin with, they might just as easily have gotten better at pretending not to be, and at giving people the answers they wanted to hear—especially if dealing with a clinic like Zucker's—without having meaningfully changed otherwise. They might have repressed their cross-gender tendencies, perhaps convincing even themselves, at least for a time, that they were better off that way, yet still been just as dysphoric as ever on the inside. They could easily have been faced with enough hostility that they went into outright denial as a survival mechanism. Anecdotally, at least, such experiences seem to be fairly common among trans people. Olivia K. Maynard, for example, reports that she "compartmentalized myself very rigidly" in response to facing "violence for being a sissy" from family members, and kept her identity buried until she was nearly forty. Similarly, activist Julia Serano "experienced intense gender dysphoria as a child" but "learned to deeply repress those feelings in order to survive" before ultimately transitioning in her early thirties. In any case, all that can honestly be claimed about this group is that they weren't openly dysphoric and hadn't undergone a transition at that time. That isn't enough to conclude desistance in any meaningful sense.
All that is little more than nitpicking, though, next to a more fundamental problem: the way the studies decided which children were transgender. Even the most recent of the studies that are cited as supporting desistance used the criteria in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) for assessing gender identity disorder in children. With the release of the fifth edition (DSM-5) in 2013, these criteria have been officially rendered invalid (though they had come under criticism long before then), and the gender identity disorder classification has itself been discarded as problematic and superseded by a new gender dysphoria classification (which emphasizes that it's the distress often associated with a cross-gender identification, not the identification itself, that needs addressing). Under the old, defunct criteria, a child would be considered to have gender identity disorder for expressing discomfort with their expected gender role, plus any four of five other criteria. Merely showing some preference for friends of the other sex and for clothing, roles, and activities typically associated with the other sex was enough to cover four of those five and lead to a diagnosis. That describes quite a few people who would never think of themselves as transgender in any sense (much less in the other-sex-identified sense being focused on here), including most tomboys, who don't generally claim to or even want to be boys. This massive loophole—which also represented a fundamental and inexplicable divergence from the adult criteria—was very deliberately closed in the DSM-5. For a childhood diagnosis of gender dysphoria in the current standard, a child must exhibit a desire or insistence to be the other sex, for its own sake and not just because of any perceived social or cultural advantages, in addition to any five of seven other criteria.
An insistent, persistent, and consistent cross-gender identification is now considered the key distinguishing trait between children who just have unexpected interests or are unsure of themselves or are playing or are going through a phase, and those who are recognizably dysphoric. Few of the subjects in question met that standard. Review of such studies has found that in the cases where the children were actually asked whether they considered themselves to be boys or girls, the vast majority, around 90%, gave answers that agreed with their assigned-at-birth sex. This disqualifies them from a diagnosis of gender dysphoria under the DSM-5. Worse, significant numbers of those included in some studies didn't even meet the then-current broader criteria of gender identity disorder in the DSM-IV. Singh's 2012 study, for instance, outright admits in section 2.1.4 that over a third of its subjects were "subthreshold" with only "some indicators" of gender identity disorder. Their inclusion is rationalized with the unsupported assertion that "some"—not necessarily all or even most—"would have met the complete DSM criteria at some point in their lives prior to assessment". That's astonishingly brazen for something that's supposed to be an objective, evidence-based, scientific study. In any case, if most subjects in these studies were never really gender dysphoric to begin with, how is it in any way remarkable that most of them didn't end up gender dysphoric and didn't go through a gender transition, which is recommended only in cases of gender dysphoria and is almost never sought out otherwise?
Children may not have as much life experience as adults, but they do tend to understand themselves better than they're often given credit for. "By age four," the American Academy of Pediatrics explains, "most children have a stable sense of their gender identity". Findings from more recent studies agree that gender dysphoric children strongly identify with their professed gender, and continue to do so, to an extent that, as one 2015 study put it, is "statistically indistinguishable from cisgender children of the same gender identity". Steensma, too, observed results in a later follow-up study that found such previously unconsidered factors as "cognitive and/or affective cross-gender identification" (emphasis added) and the intensity of early dysphoria to be good indicators of later persistence. Predicting how children are likely to turn out with regard to gender, though certainly not an exact science, is far from the blind guesswork that the desistance myth strives to portray it as. Nor is gender dysphoria something that most people can simply grow out of, or be coaxed out of. The results of early studies demonstrated not that most transgender children desist, but rather that the DSM-IV's criteria for evaluating children were flawed to an extent that made them functionally useless.
Current guidelines for care advise proceeding in ways that fit the more recent findings. Regardless of what some alarmists claim, even the easily-reversible and not particularly drastic step of social transition (which involves such things as dressing as and using the pronouns of the identified gender) is recommended only for those who meet the DSM-5 criteria for gender dysphoria, not for those who are more ambivalent or simply non-conforming. Medical intervention, on the other hand, even of the kind with no permanent effects, is not so much as considered at this age even for those who are most strongly and undeniably dysphoric. If anything, current practice may be overly cautious; one criticism is that it forces adolescents who clearly have no intention of backing down to wait longer than can be justified before starting hormone therapy, leaving them in limbo while their peers all proceed through puberty.
In any case, even for those who are not gender dysphoric, the American Psychological Association now flatly rejects efforts to force conformance as "not helpful", and the AAP similarly calls it "important to allow children to make choices" about their activities. Even if the desistance myth were true, stifling gender non-conformance in children would still be the wrong approach.
Perhaps the older studies shouldn't be blamed for their failure to make a distinction between non-conforming behavior and cross-gender identification. They used the criteria they had available, and, as far as I can tell, didn't attempt to hide their reasons for selecting the subjects that they did, even when the reasons were as flimsy as Singh's for including subthreshold subjects. However, when attempting to point to a study as a demonstration of whether children with cross-gender identification persist or desist in having cross-gender identification, it's critical that the study starts with children who actually have cross-gender identification, or at the very least that it clearly distinguishes between those who do and those who do not. Otherwise, it makes as much sense as claiming that a tree once observed as an apple seedling, as opposed perhaps to an orange seedling, later desisted from becoming a Granny Smith tree when it grew up to produce Golden Delicious apples. How can anyone either desist from or persist in something that never applied to them in the first place?
Any one of these factors on its own would make the results suspect at best. Combined, they make the results almost completely worthless. Here's what's left: After observing a small, and likely unrepresentative, sample of children, some unknown number of whom may or may not have been gender dysphoric, then making some limited attempt to follow up on them at a single later point in time, the studies judged only a fairly small portion of them to have persisted, but there were a sizable number of subjects who just never came back and so there's no telling what happened to them, and the reasons for apparent desistance in the rest went undetermined, as did whether they might go on to transition or otherwise discredit their classification as desisters in the future. There aren't many valid conclusions one can draw from such flawed data—except, perhaps, that there are people, too many to ignore, who do persist.
Note also that these results indicate nothing at all about whether transgender adolescents desist in adulthood, as that wasn't the question addressed.
The desistance myth is worthless as an argument and worse than useless for dealing with children (much less adolescents or adults) or determining policy of any kind—unless the policy involves the responsible use of data in its proper context!
 Vocabulary can get fairly problematic in this sort of discussion. "Transgender" is generally treated as a broad umbrella term that, roughly speaking, involves a disconnect between designated sex and experienced gender, and encompasses a wide variety of related ways to be human. It's a fuzzy category with no clear boundaries and no objective tests that can determine definitively whether any given person falls in or out of it, and there's some disagreement about whether certain subgroups, such as people who might not experience pronounced dysphoria as such, or who "only" crossdress or are otherwise gender non-conforming, should be included or not (the American Psychological Association's official position is essentially "yes, unless they don't want to be included"). Accordingly, credible studies (or even, in most cases, less credible studies) will be careful to explain what specific criteria they use for classification. Conversely, arguments against transgender rights are at least as likely as not to encourage a lack of clarity, especially when doing so makes their position seem stronger than it actually is. Improperly conflating distinguishably different populations is, in fact, one of the critical failings in the desistance claim, as I go on to explain.
(Discussions of sexuality, a somewhat related though fundamentally different topic, have similar problems with ambiguity, which is why, for example, the Kinsey Reports do not say about 10% of the men he studied that they were gay, but rather that their sexual behavior was "more or less exclusively homosexual for at least three years between the ages of 16 and 55". I wouldn't doubt there are men who fit that description yet would insist that they are totally straight and that it was just a passing fling that didn't really mean anything. And if they want to call themselves heterosexual, well, that's fine by me, but it's no more helpful when it comes to objective discussion and analysis than calling them homosexual would be.)
In any case, studies of desistance are typically concerned specifically with the flavor of transgender found in presumed boys who regard themselves as girls and presumed girls who regard themselves as boys, using the clinical definitions of gender identity disorder or gender dysphoria to pick them out. As such, this is the type of transgender identity that this piece primarily refers to. Though it might be more precise in this context, I hesitate to use the term "transsexual" due in part to its emphasis on physical transition, which doesn't normally apply to children and isn't necessarily pursued even in adulthood by everyone who is this flavor of transgender. On the other hand, rephrasing the desistance argument as "80% of children who may be transgender in some sense don't specifically end up as transsexual adults" helps highlight the false equivalence at its core. It's no more insightful or meaningful a claim than pointing out that most children who toss a football around don't grow up to become quarterbacks, or that most children who prefer salad to steak don't end up as vegans.
 With a sample size of ten, the margin of error is approximately ±31 percentage points when using the 95% confidence level that is the general standard. In other words, if half of the subjects had a certain result, the only conclusion we could make with 95% certainty is that this same result would apply to somewhere between 19% and 81% of the overall population that the sample represents. That isn't much better than having no data at all! For comparison, a ±10 margin of error requires a sample size of almost one hundred, and the ±3 margin of error that surveys often favor calls for over a thousand people.
 Pruning from the results those who never returned gives us a sample size of 29. That leads to a margin of error of about ±18 percentage points. Since about 72% of this smaller group was deemed to have desisted, this means that the results should properly be reported as a 54% to 90% desistance rate with a 95% confidence level, which is far less definitive and quite likely substantially lower (though potentially slightly higher) than the 84% figure that the study is generally claimed to have demonstrated. And remember, that difference comes just by addressing only the issues caused by small sample size and unfounded assumptions about those lost to follow-up, while leaving other, almost certainly larger, problems still in play, most notably the questionable inclusion criteria (as explained later in this piece).
 As earlier portions of this piece explain, the data, on the whole, is too poor to make any meaningful conclusions. However, there may be a hint of something interesting here. If 90% of the subjects identified with their assigned-at-birth sex, then just 10% of them, at most, could have qualified for a childhood diagnosis of gender dysphoria. Yet it's only 80% desistance that is claimed, which would mean that 20%—twice as many as could be diagnosed with gender dysphoria as children—did end up with a transgender identification. In other words, it seems as though quite a few more people end up transgender as adults than are discernibly transgender as children, the opposite of what the desistance myth claims! Which should come as no surprise, considering that it's often puberty that brings gender dysphoria to the fore, and that many trans people just don't understand what they're going through, or hide or repress it, until decades later. That's why it's better availability of information and improved acceptance that will do the most good, not denials.
Additional references, which are also good for further reading, many of them full of further links:
- Detransition, Desistance, and Disinformation: A Guide for Understanding Transgender Children Debates
- The pernicious junk science stalking trans kids
- The End of the Desistance Myth
- Are Parents Rushing to Turn Their Boys Into Girls?
- Methodological Questions in Childhood Gender Identity 'Desistence' Research
- Gender nonconforming youth: current perspectives