A groundbreaking analysis of four detransition types
This post breaks down the findings of one of the most rigorous data-driven studies to date on detransition, gender identity fluidity, and transition regret
A long awaited article has just been published in the Archives of Sexual Behavior: “A Latent Class Analysis of Interrupted Gender Transitions and Detransitions in the USA and Canada.”
In this first-of-its-kind project, the DARE study team used rigorous, community-engaged, data-driven methods to better understand one of the most misunderstood topics in gender medicine and LGBTQ+ politics today: detransition, identity fluidity, and gender transition regret. Data were collected between December 2023 and April 2024.
Some of the analysis shown later in this post has already presented at the last World Professional Association of Transgender Health symposium in Lisbon, at the 2024 Endocrine Society meeting, and at the 2024 Pediatric Endocrine Society meeting. It was also discussed within a guest essay published by the New York Times a few months ago, contextualizing some of the ways that the Trump administration has narrowed in on detransition in its policy moves.
But the DARE study is a complicated, highly technical project examining a complex and understudied socio-medical issue. This post will explain more about how the study was conducted, focusing on the latent class analysis element of it. Toward the end of the post we share some infographics summarizing the results.
In our view, the DARE study is one of the most rigorous studies aimed at deeply and empirically understanding detransition. Yet, because it relies on cross-sectional data gathered in the U.S. and Canada, it should be repeated by other domestic and international research teams.
About the DARE study
DARE stands for Detransition Analysis, Representation, and Exploration. The study aimed to understand different pathways to detransition, and it followed from prior insights gleaned from the Re/DeTrans Canada study. The Re/DeTrans Canada study was a qualitative pilot project that found that detransition and regret were complex, multi-dimensional phenomena that required a much more comprehensive understanding.
Given that it’s a high-stakes outcome in clinical care delivery, and can carry political implications to LGBTQ+ people’s lives, we wanted to understand more. So we collected data about these divergent experiences from people who had actual lived experiences.
The vast majority of the study participants were LGBTQ+, with 79% being assigned female at birth. The sample had an average age of 26, and many participants first realized a transgender or gender diverse (TGD) identity around age 15, give or take a few years.
Using latent class analysis of survey data from 957 participants across the U.S. and Canada, we identified four distinct classes of detransition experiences, showing that these pathways are multidimensional and complex.
What is latent class analysis (LCA)?
To inform gender care providers and to build more comprehensive, community-engaged knowledge, we leveraged latent class analysis (LCA) methodology to better identify unique, and often hidden, experiences within the heterogenous label of detransition.
LCA is a statistical method used to identify hidden sub-groups in a dataset when the groups themselves are not directly measured or pre-determined before collecting the data. LCA uses indicator variables to guide the analysis. How participants respond to specific indicator variable questions serves to identify the latent classes (subgroups).
The process looks something like this:
For the DARE study, we used participants’ self-reported reasons to stop/reverse their gender transition as indicator variables to guide the analysis.
On the survey, participants were shown questions asking them about why they “stopped transitioning, or detransitioned,” asking them to rate the extent to which psychological reasons, physical reasons, external reasons, or social reasons contributed to their decision. They were shown 21 different reasons, and they could select multiple factors (the indicator variables).
How participants responded to those 21 questions were treated as predictors of which latent class (subgroup) a participant was likeliest to belong to.
The LCA identified four different groups. We also calculated the prevalence of each type of detransition within the dataset using percentages. Class A, followed closely by Class D, contained the greatest number of survey participants.
LCA also calculates the probability that participants belong to one of the classes. So while some participants had a 0.99 probability of belonging to Classes A, B, C, or D, others were sitting around a 0.7 probability and could be considered something like border cases.
It’s important to point out that we did not create these six detransition dimensions at random. We created the list of 21 possible reasons to detransition (shown on the survey), from prior research findings, making the survey design data-driven. Here are the prior studies that we built from, and the research underlying each of these six dimensions:
Gender minority stressors (external, involuntary: discrimination, lack of support, romantic rejection. See Littman, 2021; MacKinnon et al., 2023; Narayan et al., 2021; Turban et al., 2021; Vandenbussche, 2022; Wiepjes et al., 2018).
Access to care barriers (external, involuntary: cost barriers to access desired treatments, loss of access to hormonal treatment. See MacKinnon et al., 2023; MacKinnon et al., 2024).
Neurodivergence & mental health (internal, psychological: worsening mental health, mental health complexity, realizing gender dysphoria was related to something specific. See Cohen et al., 2022; Hall et al., 2021; Littman 2021; MacKinnon et al., 2023; MacKinnon et al., 2024; Narayan et al., 2021; Pullen Sansfaçon et al., 2023; Turban et al., 2021; Vandenbussche, 2022).
Treatment dissatisfaction/medical complications (internal, medical: dissatisfaction with treatment, medical complications, health concerns. See MacKinnon et al., 2023; MacKinnon et al., 2024; Narayan et al., 2021; Vandenbussche, 2022).
Satisfaction with medical treatments (internal, medical. See MacKinnon et al., 2023; MacKinnon et al., 2024; Pullen Sansfaçon et al., 2023).
Shifting self-identity/resolution of dysphoria (internal, voluntary: re-conceptualizing gender norms, identity shift, resolution of dysphoria. See Boyd et al., 2021; Cohen et al., 2022; Littman 2021; Littman et al. 2023; MacKinnon et al., 2023; Narayan et al., 2021; Pullen Sansfaçon et al., 2023; Vandenbussche, 2022; Wiepjes et al., 2018).
But there’s even more to it. Each of these six dimensions visualized on the radar plots above corresponds to questions on the survey.
Gender minority stressors (external factors, involuntary)
a. I felt discriminated against
b. I did not have enough support in my life to continue transitioning
c. I continued to be perceived as transgender (i.e. I did not “pass”)
d. I experienced rejection from prospective romantic/sexual partners
Access to care barriers (external, involuntary)
a. I had trouble paying for hormones or surgeries
b. Legislative bans on gender care required me to stop transitioning
c. I lost access to healthcare or insurance coverage
d. I lost my housing and there was too much instability in my life to continue transitioning
e. My healthcare provider encouraged me to address my gender dysphoria with non-medical treatment options
Neurodivergence & mental health (internal, psychological)
a. My mental health did not improve while transitioning
b. My mental health was worse while transitioning
c. I discovered that my gender dysphoria was caused by something specific (i.e. trauma, abuse, autism)
Treatment dissatisfaction/medical complications (internal, medical)
a. I was dissatisfied by the physical results of the medical interventions
b. My physical health was worse while transitioning
c. I had medical complications from the medical interventions
d. I felt the changes from hormones/surgery were not enough to “pass” consistently
Satisfaction with medical treatments
I was satisfied with the physical results of the medical interventions
Shifting self-identity/resolution of dysphoria
a. My personal definition of woman or man changed and I became more comfortable with my birth sex
b. My identity changed and I no longer felt a need for medical interventions
c. I realized that my desire to transition was erotically motivated
d. My gender dysphoria resolved over time
Qualitatively testing the LCA with study participants
Once we had the LCA component of the study completed, we then interviewed a number of representatives from each of the four classes to learn more about their life experiences. We interviewed a total of 42 people in this second, qualitative phase of the study. (These were people who had already taken the survey.)
As far as we know, this qualitative feedback and test phase of the LCA results has never been done before. It’s an innovative way of using mixed-methods, and we hope to publish on our methods in the coming years.
During this qualitative interview phase of the project, at the end of each 1-1 interview, we showed participants the radar plot image of each of the four groups:
Participants were then asked to provide feedback on the dimensions, and we asked them if they could see themselves reflected in any one of the groups (either A, B, C, or D; participants were not told which group they were typed into per the LCA).
In a majority of the interviews, participants could correctly identify themselves as belonging to either Class A, B, C, or D. Interestingly, those with the highest probabilities (e.g., 0.99) could more easily tell us which group they were in, whereas those with lower probabilities (e.g., around 0.7) sometimes identified themselves in the incorrect group. (As part of this exercise, participants also shared excellent and helpful feedback on the LCA image and we hope to write a paper about this in the future.)
Once we established the groups, we were then able to use different statistical tests to characterize and compare them based on key demographics, life experiences, transition-related care received, interactions with the gender care system, and so forth.
While the four groups had many, many shared life experiences, there were also aspects of their lives that made them stand out from each other. The following infographics were created to showcase similarities and how the types of detransition stood out from each other.
Any questions related to the study may be directed to the principal investigator, Kinnon R. MacKinnon.
Thank you for reading The One Percent!
If you enjoyed this post, please share it with your friends, family, or colleagues:
You can also subscribe for free to receive new posts and support our work:
 | A guest post by
|
That is really impressive methodology and it will encourage citation. The infographics are so well designed too. Thanks for sharing!
The 80% AFAB response rate is really interesting to me. Do you think this comes down to a bias of populations surveyed (i.e., for whatever reason that AFAB people were more likely to respond), that transfeminine people are less likely to detransition (and that when they do, they're category D, with a high likelihood of retransitioning), or that the relative higher barrier / social cost for seeking transition among AMAB populations means that transfeminity self-selects for people more likely to stick with transition?
Really interesting study any way you cut it.