The frequently cited claim that 60% to 90% of transgender and gender-diverse children and young adults ultimately identify as cisgender – or their gender assigned at birth – is not supported by statistical analyses of published scientific research, according to a new study from Virginia Commonwealth University. The study, which analyzed 11 studies compiled in a commonly referenced 2016 blog post as well as five more recent publications, instead found that almost any stance on gender-affirming care for minors could be supported by different statistical analyses of the same data.
"We should rely on accuracy with our science, and we should rely on accurate science to guide legislation," said Catherine Wall , Ph.D., an assistant professor of psychology in the College of Humanities and Sciences and the study's lead author. Wall and her coauthors published their study in the journal APA PsycNet.
The quantitative meta-analysis found that "desistance" – or renouncement – of transgender identity by youth could be estimated to be as low as zero percent to as high as 100%, depending on how the studies' data were interpreted. Meanwhile, rates of persistence of transgender identity could also range from zero percent to 100%, depending on how the studies – many of which were conducted pre-1990 on small sample sizes – were analyzed.
Opponents to gender-affirming care for transgender minors referenced the original 11 studies in arguments leading to the 2025 United States v. Skrmetti decision, which ruled that state-level bans on gender-affirming care for minors were not unconstitutional. But the results of this study, Wall said, disprove the statistical foundation for the narrative that the majority of transgender youth desist from that identity.
"We were particularly interested in examining this idea of the concept of trans identity desistance because it's a number that has been bandied about for years as if it's 100% accurate," Wall said. "And we wanted to take a deeper look, especially because it's currently being used for things like legislation and bans of best-practice gender-affirming care in 26 states."
Evolving science influences data
Many of the studies listed in the 2016 blog post were conducted before 1990, when popular and scientific opinions on transgender identities were radically different. In that time, transgender identities were seen by the medical community as a pathological mental illness, and not as an inherent identity.
That outlook influenced how the studies were conducted. Of the seven pre-1990 studies, three examined effeminate behavior in young boys, and never directly focused on gender identity. Youth in the remaining four studies did not explicitly identify as transgender – instead, they were children that exhibited gender nonconforming or gender-atypical behaviors.
"Right away, we've got some significant potential issues here, where people are being referred to psychiatric care for not being a typical boy or not being a typical girl," Wall said. "They're being referred to care because of how their actions are perceived within a societal context, rather than an explicit cross-gender identification."
Additionally, in several early studies, the aim of the studies' psychologists was to counsel youth to conform to the social behaviors of the genders they were assigned at birth, now often referred to as conversion therapy. In one 1987 study, researchers separated transgender youth from their families for up to six months in order to encourage them to adapt to their assigned gender.
"A lot of those early frameworks were explicitly intending to stop people from expressing their gender identity," Wall said. "It's this idea of transgender identity as a state of disordered being."
Most of the studies also had small sample sizes, counted participants' nonresponse or discontinuation of the study as desistance from transgender identity, or primarily focused on the subjects' psychiatric concerns – such as anxiety and depression – over concerns related to their gender identity.
The researchers also noted that transgender desistance and persistence are still not well-defined, and youth may ultimately settle on a related identity, such as genderqueer or nonbinary, which could be classified as desistance in some studies.
Deeper data analysis
In this research, Wall and her coauthors performed qualitative analyses, statistical meta-analyses and simulation methods to analyze the data from the 11 studies. The researchers additionally reviewed scientific literature published since 2016 and found five related studies, which are also included in their analyses.
The researchers closely examined the limitations of the past studies, such as small sample sets. Notably, all but two studies included participants who did not respond to study follow-up assessments, and were often counted as desisting from a transgender identity.
"People can drop out of studies for any reason. Making an automatic assumption that they dropped out because they desisted in their identity isn't a fair assumption to make," Wall said.
To overcome that limitation, the researchers examined transgender persistence and desistance as separate outcomes, and ran analyses under three separate assumptions: participants left treatment for reasons unrelated to the treatment, such as moving to a new city; participants left treatment because it had a positive outcome and was no longer necessary; and participants left treatment because it had a negative outcome.
Under those assumptions, the researchers then ran analyses that either assumed that nonrespondents persisted with their transgender identities, or that nonrespondents desisted from their transgender identities. They also ran sensitivity analyses to determine how likely those outcomes were, as well as simulations to find out how volunteer bias might impact study samples.
A 'far from stable' dataset
Wall and her coauthors found that plausible rates of both desistance and persistence could range from zero percent to 100%, depending on how the 16 studies were statistically interpreted.
"We are ultimately finding that the real rate of desistance spans almost the entire range of percentages based off of different sets of assumptions," she said.
Rates of persistence also increased by nearly 5% with the addition of each study since 2016, which were generally conducted under modern scientific assumptions about transgender identities and with more precise methodologies than earlier studies, such as by only including individuals who explicitly expressed a transgender identity.
The authors assert that while some children and young adults likely do desist from their transgender identities, both the pre- and post-2016 datasets should not be used to create gender-affirming care legislation for minors.
"Understanding the context in which research is done is incredibly important to understanding the results," Wall said. "Just because a number exists does not mean it's correct, and does not mean it's 100% accurate. We should not be making legislation based off a number that might range from zero percent to 100%."
