TLDR; The Bogaert et all (2017) study which claims to find evidence that a particular protein and corresponding antibodies contribute to the FBOE has a significant problem which contradicts the main claim.

(I had investigated the below a few years ago but didn’t publish anything at the time. Scott referenced the NLGN4Y study in his recent post and I thought people might find this an interesting addition on the topic. I am not sure I’m not missing or misunderstanding something but I present it to the best of my understanding. Corrections strongly welcome.)

Hypothesis summary

The fraternal birth order effect (FBOE) is the effect that having more older brothers increases the likelihood that a later born son will be homosexual. Having older sisters may have an additional effect. 

There is some direct and indirect evidence that the effect occurs prenatally. 

One of the theories which has been proposed for this is that during a mother’s pregnancy with older brothers she develops an immune reaction to male foetus proteins which effects the development of later male foetuses (“Maternal Immune Hypothesis”, MIH).

Claim

Bogaert et all (2017) investigated a couple of Y-linked proteins to test whether one of them might be the cause of the FBOE. They claim promising results for one protein in particular.

… after statistically controlling for number of pregnancies, mothers of gay sons, particularly those with older brothers, had significantly higher anti-NLGN4Y levels than did the control samples of women, including mothers of heterosexual sons. The results suggest an association between a maternal immune response to NLGN4Y and subsequent sexual orientation in male offspring.

And

In line with the maternal immune hypothesis, the predicted ordering for antibody concentrations to anti-NLGN4Y was as follows: women with no sons < mothers of heterosexual sons < mothers of gay sons with no older brothers < mothers of gay sons with older brothers.

Initially this result seems promising – a higher level of anti-NLGN4Y in mothers seems to correlate with increased likelihood of homosexuality, even after controlling for number of pregnancies.

Problem

However there is a big problem – the data indicate that having more pregnancies actually decreases the level of antibodies produced by mothers. 

As shown in Table S4, the correlations [between number of pregnancies and antibody concentrations] were negative in direction and were generally significant for the three anti-NLGN4Y measures, ranging from −0.153 to −0.229. These results suggested that pregnancy per se might have a tolerogenic effect (39) with regard to NLGN4Y.

This is completely the opposite of the prediction by MIH. 

(I note that this correlation is for all pregnancies, not just male foetuses. Even if FBOE only applies to males, MIH wouldn’t predict the correlation for all pregnancies to be negative unless having older sisters made later sons strongly less likely to be homosexual.)

This negative correlation is even more problematic than it appears - the analysis is performed “after statistically controlling for number of pregnancies”. This controlling will (relatively) increase the amount of antibodies in women with more previous pregnancies – the mothers who we already know from FBOE are more likely to have homosexual sons. This suggests that the positive results found for the headline claim are, at least in part, caused by the controlling for number of pregnancies.

Why not addressed?

The paper does not even address this issue – it doesn’t seem to notice that this is an issue for the hypothesis.

In an earlier paper, Bogaert and Skorska (2011) list necessary conditions for the MIH (maternal immune hypothesis):

A fifth condition for the MIH to be a plausible explanation of FBO is that the maternal immune response to Y-linked proteins should show an incremental response to previous male foetuses similar to the incremental pattern that the FBO effect has on male sexual orientation for each additional older brother.

So the lead authors are aware that this result is not as would be predicted.

I don’t want to be overly harsh here but I can’t help feeling that if the correlation had been in the direction predicted by the theory then this result would be in the abstract and the results and the discussion, not hidden away in the very last paragraph of the paper.