Brain Sex in Transgender Women Is Shifted towards Gender Identity

https://www.mdpi.com/2077-0383/11/6/1582

A short critique of the paper:

https://xcancel.com/FondOfBeetles/status/2077389417291845705

I’ve requested the subject-level Brain Sex Index data behind Kurth et al., 2022. Brain sex in transgender women is shifted towards gender identity. JCM.

While I wait, some open questions on the published methods and conclusions.

---

(Using the paper language)

The key comparison (transgender women vs. cisgender men) was “one-tailed”(seeking a statistical difference in only one direction, which provides a more generous boundary to reach significance).

There is no stated correction for multiple pairwise comparisons across multiple groups.

Two-tailed, their own reported p=0.016 becomes 0.032. Why one-tailed, and why no correction across the 3 group comparisons?

---

The cisgender men’s group has one data point flagged as a statistical outlier (marked “+” on the plot, and apparently excluded from the box’s standard stats display).

But the reported comparison test shows 24 cisgender men were included.

If one cisgender man is such a clear outlier, does the headline finding hold with that point removed?

---

Effect sizes (d): briefly, how large a difference are we looking at?

Transgender women vs. cisgender men, d=0.64 (moderate).
Transgender women vs. cisgender women, d=1.87 (large).

That’s a big asymmetry: is that gap large enough to justify the authors’ conclusion of “hovering between”.

And since this calculation includes the outlying man, would removing it make “hovering between” a nonsensical conclusion?

---

As an analogy: the cisgender men are at the 100m finish line, the transgender women are about a 25m behind them, and the cisgender women are still at the start.

---

The authors report their transgender women cohort (n=24) as 6 androphilic (attracted to men) and 18 gynephilic (attracted to women).

Orientation has been linked to brain differences in other studies.

Why wasn’t orientation controlled for or stratified, especially since the authors recommend this for future studies? 😂

You have the data, my friends.

What does their dataset look like when disaggregated for orientation?

---

Anyway, I’ve made a formal request; I know other researchers have too. I’ll share what I get.

(And again, I’m repeating the language of the paper for ease of cross reference to it)

---

Frankly, to make the claim that sexuality might impact analysis in TIMs, look at the TIM plot that clearly shows two groups - as opposed to the smooth bubble of other males and the female cohort - and fail to disaggregate is bonkers.

There is a reason this is a different shape.