Falsification of GID prevalence results by the APA Task Force on Gender Identity and Gender Variance
An Investigative Report by Lynn Conway
28 August 2008 [V 8-29-08] [PDF]
See also Open letter of 9-05-08 to the President of the APA [PDF]
The Report of the APA Task Force on Gender Identity and Gender Variance [APA08] grossly underreports the prevalence of “gender identity disorder” by a factor on the order of 10 to 20.
The underreporting results from a misrepresentation of the results of [Bakker93], failure to report error bounds on [Bakker93] (as discussed in [WPATH01]), failure to report errors in the calculations in [Bakker93] (as revealed in [Olyslager&Conway07]), and failure to report counter-evidence and results of recent prevalence studies [Olyslager&Conway07, Winter02, Winter06].
The Task Force’s misrepresentation of GID prevalence is worsened by their reporting of results to three significant figures, incorrectly suggesting that the reported numbers are absolute ones (instead of lower bounds) and that they are accurate (+/-) to within a very small percentage.
Falsification of the results of [Bakker93]:
We find the following statement on pages 37-38 of [APA08]:
“Prevalence4. Based on referrals to a national, government-subsidized gender identity clinic in the Netherlands, the prevalence of gender identity disorder in adults was estimated to be 1:11,900 for male-to-female transsexuals and 1:30,400 for female-to-male transsexuals (Bakker, van Kesteren, Gooren, & Bezemer, 1993).”
That statement falsifies the results of Bakker, et al, by asserting that the numbers 1:11,900 and 1:30,400 in [Bakker93] are estimates of the “prevalence of gender identity disorder”. In reality, Bakker, et al estimated the prevalence of “sex reassignments” in the Netherlands in the late 80’s, as clearly discussed in the Abstract of [Bakker93]:
“The prevalence of transsexualism in the Netherlands was estimated by counting all the subjects who were diagnosed as transsexual by psychiatrists or psychologists and were subsequently hormonally treated and generally underwent sex-reassignment surgery.”
For readers’ convenience, we’ve posted a scan of [Bakker93] on the net. Readers can easily confirm that [Bakker93] reports sex reassignment data rather than GID data.
How this happened and the effect it has:
Bakker, et al, followed the habit of European researchers in the 70’s, 80’s and 90’s of referring to the prevalence of “sex reassignment” as being “the prevalence of transsexualism”, which it is not.
This practice was a careless shift away from the paradigm established by Wĺlinder [Wĺlinder68], in which he attempted to estimate the prevalence of the underlying condition of intense gender dysphoria, which he defined as ‘transsexualism’. However, SRS’s are easier to count than are people with gender dysphoria, so SRS’s are what later European researchers counted.
The shift in meaning in [Bakker93] led to confusion for many decades. After all, most people want an answer to the question “How likely is it that someone might experience gender dysphoria?” The far smaller counts of “sex reassignments” answer a different question.
Key members of the APA Task Force were well aware of the true meaning of the [Bakker93] results, because Olyslager and Conway had exposed it in a presentation at the WPATH 2007 Symposium [Olyslager&Conway07]. In particular, members Zucker and Lawrence (WPATH’s experts in GID prevalence and responsible for revising that section in the 7th Ed. of the SOC) were quite familiar with [Bakker93] and with its deconstruction by [Olyslager&Conway07].
It thus appears that the Task Force knowingly misrepresented the results of [Bakker93] by referring to “sex reassignment” numbers as numbers for “gender identity disorder” – thereby making intense gender dysphoria appear to be far less prevalent than it actually is.
Failure to mention known sources of estimation errors:
The Task Force was also well-aware of this statement in WPATH’s Standards of Care, 6th Ed:
“Prevalence. When the gender identity disorders first came to professional attention, clinical
perspectives were largely focused on how to identify candidates for sex reassignment surgery. As
the field matured, professionals recognized that some persons with bona fide gender identity
disorders neither desired nor were candidates for sex reassignment surgery. The earliest
estimates of prevalence for transsexualism in adults were 1 in 37,000 males and 1 in 107,000
females. The most recent prevalence information from the Netherlands for the transsexual end of
the gender identity disorder spectrum is 1 in 11,900 males and 1 in 30,400 females. Four
observations, not yet firmly supported by systematic study, increase the likelihood of an even
higher prevalence: 1) unrecognized gender problems are occasionally diagnosed when patients
are seen with anxiety, depression, bipolar disorder, conduct disorder, substance abuse,
dissociative identity disorders, borderline personality disorder, other sexual disorders and
intersexed conditions; 2) some nonpatient male transvestites, female impersonators, transgender
people, and male and female homosexuals may have a form of gender identity disorder; 3) the
intensity of some persons' gender identity disorders fluctuates below and above a clinical
threshold; 4) gender variance among female-bodied individuals tends to be relatively invisible to
the culture, particularly to mental health professionals and scientists.” [WPATH01]
However, when reporting the very same [Bakker93] results as had [WPATH01], they did not mention the above sources of estimation error discussed long before in [WPATH01]. They instead reported their results to three significant figures, suggesting that the numbers are absolute ones (instead of lower bounds) and that they are accurate (+/-) to within a very small percentage.
Failure to mention calculation errors in [Bakker93]:
To make matters worse, the methods used in [Bakker93] to calculate the prevalence of sex reassignments were erroneous, as also revealed in [Olyslager&Conway07]. If correct calculations had been applied to their data, they would have reported an MtF “sex reassignment” prevalence of about 1:3,500 and FtM of about 1:6,200. The APA Task Force failed to mention these well known errors in [Bakker93], and reported the incorrect 1:11,900 and 1:30,400 numbers instead.
If we take both the shift in meaning and the calculation errors in [Bakker93] into account, we find that Bakker et al’s own data implied a lower bound on the prevalence of MtF “gender identity disorder” in the range of 1:1,000 to 1:2,000, as determined in [Olyslager&Conway07], a full order of magnitude higher in prevalence.
Failure to mention counter-evidence and open dismissal of counter-evidence:
Early research in Singapore found a prevalence of MtF sex reassignment of 1:2,900 [Tsoi88], implying a prevalence of gender dysphoria of 1:1,000 or more. Similar values were projected for the U.S. in [Conway01]. Recent studies in Thailand indicate a prevalence of MtF gender dysphoria of 1:500 or more [Winter02, Winter06, Olyslager&Conway07], a level at least 20 times that reported by the APA. The APA Report fails to mention or cite any of these studies.
The APA Report does mention the Olyslager and Conway report [Olyslager&Conway07] in a footnote, but dismisses it with the following snide comments [APA08]:
4Olyslager and Conway (2007) suggest that the figures cited here are low. This paper, however, seems to represent a minority position among researchers, although transgender activists tend to endorse the study.
Those comments fail to reveal that Olyslager and Conway had exposed systematic definitional and mathematical flaws in all earlier European GID prevalence studies (including [Bakker93]) – flaws that had led earlier work to greatly underestimate the prevalence of gender dysphoria.
They instead dismiss Olyslager’s and Conway’s analysis by claiming it fails a majority vote test when put up against a count of those earlier European papers – rather than evaluating the analysis on its own merit.
That dismissal is scientifically improper, especially since no challenge has been made to the calculations, logic and results in [Olyslager&Conway07].
The APA Report’s footnote 4 also gives the false impression that [Olyslager&Conway07] is included in the Report’s references. However, the Task Force failed to list the full citation for [Olyslager&Conway07] in the references, effectively suppressing it, so interested parties cannot confirm the counter-evidence themselves.
Summary and Findings:
The APA Task Force Report on Gender Identity and Gender Variance [APA08] greatly underreports the prevalence of “gender identity disorder” by a factor on the order of 10 to 20.
The underreporting of GID prevalence derives from a misrepresentation of clinical definitions and a failure to mention known calculation errors in sources.
The unreasonably low prevalence numbers are given to three significant figures in the Report, as if they were precisely accurate – while failing to mention well-known sources of estimation error.
The Task Force then dismisses recent work by Olyslager and Conway that had exposed large errors in earlier studies by calling that work a “minority position” – as if a scientific analysis must be certified by a majority vote, rather than judged on its merits.
The Task Force further dismisses the work of Olyslager and Conway by insinuating that citation by "transgender activists" somehow reduces its validity – while failing to cite it themselves.
Finally, the Task Force fails to mention recent scientific studies that report far higher-levels of GID prevalence than does their Report.
[APA08] APA Task Force on Gender Identity and Gender Variance. (2008). Report of the Task Force onGender Identity and Gender Variance. Washington, DC: American Psychological Association. Note: This publication was posted on the internet and widely disseminated as a PDF file on August 17, 2008, with considerable fanfare in the press [APA08a].
[APA08a] “APA resolves to play leading role in improving treatment for gender-variant people”, APA press release, August 17, 2008.
[Bakker93] A. Bakker, P.J.M. van Kesteren, L.J.G. Gooren, and P.D. Bezemer, "The prevalence
of transsexualism in the Netherlands," Acta Psychiatrica Scandinavica, vol.87, pp.237-238, 1993.
Mirrored on: http://ai.eecs.umich.edu/people/conway/TS/Prevalence/Reports/REFs/bakker1993.pdf
[Conway01] L. Conway, "How Frequently Does Transsexualism Occur?", LynnConway.com,
Jan. 30, 2001.
[Olyslager&Conway07] F. Olyslager, F. and L. Conway, “On the Calculation of the Prevalence of Transsexualism”, presented at the WPATH 20th International Symposium, Chicago, Illinois, September 6, 2007. (Submitted for publication in the International Journal of Transgenderism).
[Tsoi88] W.F. Tsoi, "The prevalence of transsexualism in Singapore," Acta Psychiatrica
Scandinavica, vol. 78, pp.501-504, 1988.
Mirrored on: http://ai.eecs.umich.edu/people/conway/TS/Prevalence/Reports/REFs/tsoi1988.pdf
[Wĺlinder68] J. Wĺlinder, Transsexualism: Definition, prevalence and sex distribution , Acta
Psychiatrica Scandinavica, vol. 203, pp.255-257, 1968.
Mirrored on: http://ai.eecs.umich.edu/people/conway/TS/Prevalence/Reports/REFs/walinder1968.pdf
[Winter02] S. Winter, "Counting Kathoey", Transgender Asia Papers, August 27, 2002.
[Winter06] S. Winter, Thai Transgenders in Focus: Demographics, Transitions and Identities,
International Journal of Transgenderism, vol. 9, No. 1, pp.15-27, 2006.
[WPATH01] W. Meyer III et.al., "Standards of Care for Gender Identity Disorders, 6th Version,"
World Professional Association for Transgender Health, Feb. 2001, p.2.
For more information also see:
“Notes to Reviewers of the paper "On the Calculation of the Prevalence of Transsexualism" By Femke Olyslager and Lynn Conway, September 6, 2007.”
Those notes contain contact information, links to tutorials on the mathematics used in that paper, links to related work in progress, and links to the originals (PDFs) and summaries of the data and results of all the key past papers - enabling readers to see those papers in original form, and thereby confirm our analyses of the data and results in those earlier papers.
LynnConway.com > TS Information > Prevalence of Transsexualism > Falsification of GID prevalence results by the APA Task Force on Gender Identity and Gender Variance