A Critique of the Autogynephilia Hypothesis

Part 1. Empirical Evidence

This article was written in response to The Man Who Would Be Queen by J.Michael Bailey.

by Catherine Anderson, PhD


The empirical studies used by Bailey and Blanchard to conclude that there are "two, and only two" types of transsexualism are examined. It is shown that these studies plainly do not permit this strong and sweeping conclusion. J. Michael Bailey's recent book [1] asserts that there are two, and only two, types of transsexuals, one of which is associated with autogynephelia (AG). His principle empirical evidence consists of a few studies of Ray Blanchard. If one reviews these studies, it becomes apparent that they do not justify the sweeping conclusions that Blanchard and Bailey reach. Personally, I believe that autogynephilia is a mental dynamic that occurs with some transsexuals and crossdressers. My objection is to the unrealistic "black or white" manner in which Blanchard and Bailey have insisted that autogynephelia is both an distinct, proven clinical entity (as opposed to simply a helpful construct), and, worse, that it is one of only two such entities. The Autogynephilia Hypothesis Briefly, Blanchard and Bailey suggest that there are two categories of male-to-female transsexuals: homosexual and nonhomosexual. Homosexual transsexuals begin life as feminine boys who, from an early age, (a) identify as being female, and (b) are more attracted to men than women. The nonhomosexual transsexuals develop a female identification later in life. Before this, their attraction is basically heterosexual. Blanchard and Bailey suggest that these individuals have AG, loosely understood as a fetishistic attraction to the idea of themselves as a woman. By constant repetition of the sexual fantasy of being female, reinforced by accompanying masturbation, the fantasy gradually becomes confused with their personal identity. In mild cases, the men remain heterosexual crossdressers. In extreme cases, they begin to take female hormones, and may progress to seek sexual reassignment surgery. Supporting Studies Blanchard [2] cites three studies as empirical evidence of the AG hypothesis. (There are other studies, but we can take these as representative.)

The reader is encouraged to review the presentation where Blanchard summarized this evidence, available as a PowerPoint presentation on Professor Bailey's website <http://www.psych.nwu.edu/psych/people/faculty/bailey/controversy.htm> [3]. Scientific Logic of Blanchard's Studies In two important respects, Blanchard's studies beg the question. First, in summarizing his research, he states:

"The taxonomic question, therefore, was not whether there is more than one type of transsexualism in males, but rather, how many more than one type, and how these should be characterized."[2]

At the outset, the arbitrary assumption is made that there are "types" of transsexuals. We all know there is heterogeneity among transsexuals. But heterogeneity per se does not imply the existence of "types."

Within psychology there is no consensus on whether human behavior is better explained in terms of types or in terms of continuous dimensions (or some combination of both). This topic, for example, has been extensively investigated in the area of depression. Are there distinct subtypes of depression (e.g., major depression vs. minor depression), or is there simply a continuous dimension of "depressedness" on which people fall? That issue is nowhere close to being resolved.

A correct way to describe Blanchard's experimental hypothesis would be this: If there are fundamental types of transsexualism, what might those types be?" This formulation makes clear that his studies offers no insight as to whether the "type" model is better than a "dimension" model.

A dimensional view might be a better way to think of AG. One would assert that there is an "autogynephilia tendency" which may be present in varying levels in different people. That a much milder statement, and more reflective of the complexities of individual differences, than to state that AG is a distinct psychological entity and constitutes one of only two possible states.

Blanchard begs the question in another significant way. He states:

"The research strategy that I used for this question was to start by distinguishing a larger number of groups and then reduce this to a smaller number by combining groups that seem to be merely superficially different variants."[2]

The "larger number of groups" is Hirschfeld's [4] distinction of four different categories of transvestite: (1) heterosexual; (2) bisexual; (3) asexual; and (4) homosexual. Blanchard relates failure to find differences between categories (1), (2), and (3) above on certain variables. On this basis he suggests we should only distinguish two types of transsexual: a homosexual and a nonhomosexual variety. (He then equates the nonhomosexual variety with his AG hypothesis).

The proper conclusion of Blanchard's study would simply be that Hirschfeld's taxonomy is poor, or that a Blanchard's two-category taxonomy is better Hirschfeld's fourfold taxonomy. What Blanchard and Bailey actually conclude is something very different: that there are two types of transsexuals--period.

The distinction between these conclusions is subtle and I shall emphasize it. It is one thing to say that the homosexual vs. nonhomosexual distinction is better than Hirschfeld's fourfold categorization. It is quite another to say that Blanchard's twofold distinction is better than all other typologies-which is what Blanchard and Bailey suggest. That is simply not the correct conclusion permitted by the studies. There may be an infinite of alternative taxonomies which are vastly superior to the homosexual vs. nonhomosexual typing. Blanchard's studies do not address this broader issue at all.

It is important to recognize the difference between a category and a type. To use a medical example, one might distinguish between "mild" and "severe" forms of flu. In a restricted, colloquial sense, one might say that these constitute two "types" of flu. What that really means is "on the attribute dimension of severity, there are two categories of flu-mild and severe." In reality we know there are many more types of flu, e.g., those caused by different viruses, those with different symptoms, etc.

The distinction between types and categories might be the basis for some of the confusion and controversy surrounding Blanchard's and Bailey's statements. I submit that a better way to summarize their ideas is to say "on the attribute dimension of 'attracted to the idea of being female' there are two categories of transsexuals-those who are and those who aren't." But, just as with the flu example, that is far from saying this distinction constitutes the best overall taxonomy to describe all variation among transsexuals.

Finally, reviewing the Blanchard's data makes it perfectly clear that his two proposed types do not account for all the results. Note that among the "nonhomosexual" samples, there is less than 100% endorsement of the items that purport to demonstrate AG tendencies, and among the homosexual samples, there is greater than 0% endorsement of these items.

Implications for Future Research and Practice

At this point, AG is best viewed as a heuristic construct, not a proven diagnostic category. An unfortunate tendency with diagnostic labels is that they function as "attractors." In numerous historical examples one can see how a new diagnosis begins to attract under its rubric more and more cases, which become increasingly unrelated to the original definition. A case in point is Attention Deficit Disorder (ADD). Originally this was proposed to describe a specific, not especially common condition, with a potential biological basis. However, over time the ADD label has come to be applied injudiciously to nearly any youth whose behavior or intellectual style deviate from the norm. A diagnostic label is often an intellectual fetish-something that provides an irrational sense of security and control to the person who uses it.

This is a very real danger in the case of AG-that it will become a diagnostic fad, and applied as a "defining" category far more often than it should. ("A man with a hammer, sees the world as nails.) This dehumanizes a client.

The real purpose of a heuristic construct like AG is to guide a deeper understanding of the individual. Knowing that such a thing as AG exists, a therapist or counselor can seek to understand how it might apply to the client, and how it might interact with other aspects of the client's personality. It is unlikely that all the person can be understood in terms of this one construct.

There does seem to be fairly wide agreement that something like what Blanchard calls AG does exist. Future discussion should shift to (1) what causes AG, when it does exist; (2) what are the underlying psychological processes associated with it; (3) what are the implications for who are good candidates for SRS and who are not; and (3) is there a subset of AG males who are better off remaining with a male gender, and for these, can psychotherapy or other modes of intervention help?

About Part 2 of this Note

As a theory, there are two major shortcomings of the AG hypothesis First, it does not distinguish between correlation and causation. Blanchard and Bailey have observed that, in some transsexuals, a temporal association exists between sexual arousal, crossdressing, and fantasies of the self as female. But, one cannot infer from this mere correlation that these people crossdress or fantasize for the purpose of arousal. The arousal could be a byproduct of the dressing or fantasies, or both the dressing or fantasies and the arousal could be due to a third factor. Second, no attempt is made to explain why some boys experience AG and other do not. We cannot exclude this possibility: that AG is due to some form of innate femaleness in certain males.

An alternative to the AG hypothesis that meets these concerns, and is equally compatible with the evidence is presented in Part 2 of this note. The hypothesis advanced is that many late-onset transsexuals have a biologically-mediated female sexuality (which is a different issue than sexual partner preference). It is also suggested that their late onset makes sense given a two-dimensional view of gender (maleness and femaleness as separate dimensions). Transsexuals with high levels of maleness may take longer to admit their transsexualism and may experience higher levels of conflict regarding it.


[1] Bailey JM. The man who would be queen. Washington D.C.: Joseph Henry Press, 2003.

[2] Blanchard R. (2000). "Autogynephilia and the taxonomy of gender identity disorders in biological males. In J. M. Bailey (Chair), Phenomenology and classification of male-to-female transsexualism. Symposium conducted at the meeting of the International Academy of Sex Research, Paris.

[3] http://www.psych.nwu.edu/psych/people/faculty/bailey/controversy.htm

[4] Hirschfeld M. Transvestites: the erotic drive to crossdress. 1910? 1925? (English translation by MA Lombardi-Nash published by Prometheus Books, UK, 1991).

[5] Blanchard R. Typology of male-to-female transsexualism. Archives of Sexual Behavior, 14, 247-261, 1985.

[6] Blanchard R. (1989). The concept of autogynephilia and the typology of male gender dysphoria. Journal of Nervous and Mental Disease, 177, 616-623, 1989.