Wednesday, June 17, 2015

Persuasion by proof or proof by persuasion?

I recently emailed an eminent psychiatrist at a world-renown teaching hospital about transgenderism. I won't say who it was–although I've shared the correspondence with a few trusted friends. This is his response. (I've edited out the personal references):


There are a few studies out and about that make confusion rife in this domain. What’s noteworthy about them all is how selective they are, how they confuse matters of intersex and well known injuries to the occasional fetus with the vast majority of transgender cases where the causes are clearly psychological, What I always ask these folks is not to tell me what “current understanding” consists of but give me a reference to study. Then as usual–being controversialists rather than clinical scientists–they write back something ad hominem such as “you’re not mainstream” “you’re not a good judge” etc. I can’t therefore comment on “this putative evidence” because it’s all so “putative.”

You may remember, the Multiple personality/False memory craze some ten years ago. Again at that time, I’d hear from fellow psychiatrists–including many “leaders”–that they were “right behind you” but worried that their reputations would be damaged if they came out for me before the win was official!! It was extremely amusing as it revealed the careerism of so many of them. I’ve seen this set of tactics before–I put it this way: Controversialists want to “prove by persuasion” and use every trick they know to “persuade” including silencing you, insulting you, dismissing you etc. I’m, at least I think of myself as, a clinical scientist who wants to “persuade by proof” and am looking to open up the arguments by proposing what counts as proof and what needs to be challenged as “non-proof.”


  1. 1. At least from what I've seen, I think the main problems in "transgender" or "transsexual" studies is selective methodology which biases the studies, confusing transsexual with intersexual or similar problems wherein transsexual issues are far and away solely psychological (e.g. Jenner), and the sample sizes of many if not most of the "studies" are way too small to derive any meaningful conclusion.

    2. For example, let's take sample size. Sample size is or can be crucial to the reliability of a study.

    All things equal, a pharmaceutical company can't necessarily test a new drug on a group of 100 people, and if the drug is effective on 67 of these 100 people, then generalize its effectiveness to the entire population of the US, and start making, marketing, and selling the drug to everyone. To all 300+ million Americans. The company needs a minimum sample size (which is normally worked out before a study even begins) in order to tell whether the effectiveness of the drug will be able to be generalized to a greater population. What that minimum sample size will be will vary depending on the specific study, its treatment or intervention plan, goals or purpose of the study, etc.

    However, if the writer readily admits from the get-go that the sample sizes are too small, and presumably not well powered, then at best the studies would have to be inconclusive one way or the other.

    3. Another lesser but hardly insignificant problem I've seen in transgender or transsexual studies is there is sometimes no control group when comparing a treatment. However, without a control group, then there's nothing to compare the intervention group to.

    Generally speaking, a good study needs at least two groups - a control group and an intervention group. If we are testing a new drug out, then we have to compare how it works in relation to an old drug. To see if it's better or worse. Or at least we need a placebo control group that doesn't the old or new drug at all. But without any control group, how much can we really conclude?

    4. Also, I've seen some argue along these lines:

    "While no single study presents proof beyond any shadow of a doubt or with metaphysical certainty, taken together they do present a preponderance of evidence such that one can say with confidence that transgender or transsexual individuals have a congenital gene-based difference from cissexual individuals."

    Quantity of studies doesn't necessarily mean anything. Quantity doesn't trump quality. We could have 100 studies arguing that the earth is flat. But if there's a single good quality study demonstrating the earth isn't flat, then that study would trump the other 100 studies.

    5. Many studies use radiological imaging technology (e.g. MRI +/- PET) to deduce "female patterns" in male-to-female transsexuals.

    Of course, there's a lot we can tell from radiological imaging. But there are also limitations. I won't go through and weigh the pros/cons of using each imaging modality, unless the occasion calls, but it's something to keep in mind for now.

    1. 6. Many studies make the elementary mistake that correlation is tantamount to causation. But we have to beware not to necessarily draw the connection.

      7. Many studies seem mainly to look at male-to-female transsexuals, but I wonder if they are looking at these male-to-female transsexuals after the fact, after they've become transsexuals via hormone replacement therapy or sex reassignment surgery? But what we really want to know is what transsexuals were like prior to undergoing their changes to become male-to-female, for at least theoretically speaking it's possible the transformation process could have somehow affected or altered their neuroanatomy or neurophysiology (e.g. androgen receptor length or number). Or have these studies properly controlled for this?

      8. Alfred Jost is the scientist most responsible for coming up with the standard model of normal sexual differentiation in humans. According to the Jost model, genetic sex determines gonadal sex, and gonadal sex determines phenotypic sex.

      In short, there are broadly speaking three main determinants of sex, the prior determining the following:

      i. Genetic sex: the presence or absence of a Y chromosome.
      ii. Gonadal sex: the presence of testes or ovaries.
      iii. Phenotypic sex: the appearance of the external genitalia.

      However, like the DNA-RNA-protein (aka the central dogma of molecular biology) there's more than meets the eye. For example, a primary player in the Josts model is the SRY gene on the Y (male) chromosome which dictates sexual differentiation. But as any reasonable physiologist, embryologist, and other relevant scientist will admit, the mechanisms by which the SRY gene dictates testicular development and testicular hormones to promote male development are at best incompletely understood.

      9. Related, we can say hormones control sexual development and differentiation to a significant degree. But what we can't say (since we simply don't know) is whether the mechanisms controlling sexual differentiation of the human brain are *only* or exclusively hormonally determined.

    2. #4 - to elaborate, I'm assuming poor quality in the 100 studies, but good quality in the single study. Of course it'd be best to have a tremendous quantity of quality studies.

    3. Good thoughts rwh. The only problem with your logic and reason is that it is logical and reasonable, which just won't do.

      As soon as the LGBT-lobby makes a gain they drop prior arguments and shift ground.

      Genetic and physiological studies aren't important anymore, what's important is "perceived self-identity". That's what matters. That's what's "authentic".



      And so it goes...