Box Turtle Bulletin

Box Turtle BulletinNews, analysis and fact-checking of anti-gay rhetoric
“Now you must raise your children up in a world where that union of man and box turtle is on the same legal footing as man and wife…”
This article can be found at:
Latest Posts

Some Concerns About Study Refuting Fraternal Birth Order Effect

Timothy Kincaid

December 1st, 2008

In January of 2008, Andrew Francis, an assistant economics professor at Emory University, released a paper in which he sought to apply analysis to the National Longitudinal Study of Adolescent Health (“Add Health”). His results are published in the October 2008 Journal of Sex Research and have gotten some attention from those who follow the nature / nurture debate.

His abstract includes the following:

For men, I find that having one older brother does not raise the likelihood of homosexuality. Although having multiple older brothers has a positive coefficient, it is not significant. Moreover, having any older sisters lowers the likelihood of homosexual or bisexual identity.

Unfortunately, I am neither a statistician nor am I able to fully comprehend his methodology or make heads or tails of his charts. I will leave that to those better skilled than I.

But I do have some observations that cause me to be reluctant to quickly accept Francis’ assertions.

Bias: I do have concern about Francis’ motivations. In August 2007 he released a paper in which he applied economic theory to sexual behavior and HIV and declared:

I present evidence that AIDS causes some men to shift from homosexual to heterosexual behavior, whereas AIDS causes some women to shift from heterosexual to homosexual behavior. Thus, sexual behavior may respond to incentives. I consider alternative hypotheses, including biological theories of sexual orientation and stigma-related survey bias, and argue that they are unlikely to explain the results.

In a paper written in March of this year, he proposed that a 1 to 2 year incarceration be the penalty for intentionally or unknowingly transmitting HIV. And while I find that paper to be troubling, he also released a paper in October that suggested that increased social tolerance towards homosexuality causes a statistical decrease in HIV infection rates.

So please note that I am not accusing Mr. Francis of bigotry; he’s not Paul Cameron.

But behind all of Mr. Francis’ work is the assumption that sexual behavior is malleable – at least on the edges – and it’s strongly implied that orientation (what Francis calls “desire”) is as well. So it should be greeted with caution that he has released a study that confirms what he has been presuming to be true for at least a year.

Contradictory: The most important question to ask of any study is, “Does this make sense”.

While it is entirely possible for ‘common knowledge’ to be completely wrong and for research to be a tool that overthrows myth and supposition, it doesn’t overthrow mathematics or logic. And Francis’ results reveal ‘findings’ that are hard to fathom.

Francis measures sexual behavior, sexual attraction, and two measures of sexual identity: not entirely heterosexual and not even close to heterosexual. But in Francis’ model these are often contradictory and in ways that are difficult to rationalize.

For example, being raised by a single dad has no material effect on sexual desire or sexual behavior but it is a positive indicator on identity. Say huh?

Yep. Boys raised by a single father are more likely to identify as gay but they aren’t attracted to the same sex or doing anything about it.

And a few other things he found… discoveries that will certainly delight some folks who have notions about classes and races.

Young black men are more attracted to guys and are having more sex, but (except for the nelly ones) they deny being anything but 100% straight. And if a boy is more educated, he’s more likely you are to declare himself gay (while remaining virginal), while blue-collar “straight” studs are out there humping like rabbits.

Frankly, that’s a worldview that is best left to porn.

Where it all falls apart for me is in the comparison of desire and identity.

I can get that there might be some demographic that expresses desire and romantic attraction to other men but cannot or will not identify as other than heterosexual. But it makes no sense whatsoever for a study to claim that there are four demographic subgroups that are more likely to identify as gay but aren’t attracted to the same sex.

Conclusion: You can measure the credibility of a research by whether their declarations of discovery are well supported by the data.

But consider this doozy:

Given the complexity of the empirical relationship between sexual orientation and the biodemographic and other correlates, it is likely that both biological and social mechanisms may play a role.

Wait. Social mechanisms? When, oh when, did the data address social mechanisms? The closest he gets is looking at college education.

But in Francis’ assumptions, if there’s no older brother effect then it’s likely that social mechanisms may play a role. Yikes.

I don’t know whether there is adequate evidence to conclusively prove whether there is a fraternal birth order effect in male sexual orientation. And I’ll let those who are more knowledgeable tell us if this model is conducted properly.

But I think I’ve seen enough to know that I’ll not place much reliance on this paper.

Comments

POST COMMENT | COMMENT RSS 2.0 | TRACKBACK URL

Andrew
December 1st, 2008 | LINK

I love reading your response to stats and studies. It’s nice knowing I’m not the only person who reads them and immediately asks the basic method questions!

Warren Throckmorton
December 1st, 2008 | LINK

Timothy – For a psych paper, there are a couple of oddities. Francis uses the 95% confidence interval and does not report that these regression coefficients are likely not significant at the 99% confidence level. He does not report effect sizes which is now standard in APA journals. The Journal of Sex Research, while fine, is not an APA journal. If he had reported effect sizes, all of the significant findings would have had effects of much less than 1%.

This does not invalidate the paper but does invite a closer look. When you look closer, you find that the predictive value of the family demographics is trivial. None of them matter much at all. The reason you get any differences is the large numbers involved given that he used the ADDHealth data.

The logical inconsistencies in the variables you noted as predictive owe to the trivial size of these differences. In essence, there is next to nothing here that helps predicts whether someone will report a same-gender partner, or romantic same-sex attraction or an identification that is a smidge away from 100% straight. A couple of blips on the screen do not make for a coherent picture. The picture is even more strange when you try to make sense of the female side of the house.

Francis study is valuable not for the variables he identified but for the confirmation that these variables have only a trivial level of predictive power. I wish he have been more clear about that. I have yet to contact him; perhaps he would disagree but that is how it looks to me.

quo III
December 1st, 2008 | LINK

“Young black men are more attracted to guys and are having more sex, but (except for the nelly ones) they deny being anything but 100% straight.”

That’s extremely interesting if it is true, because it refutes the idea, which I have seen repeated often, that the supposed fact that there is not a higher rate of homoesxuality among blacks refutes traditional Freudian theories about distant fathers as a cause of homosexuality (the reasoning is that if those theories were true, then there should be a higher rate of homosexuality among blacks because more of them grow up without fathers, but supposedly there isn’t).

Francis deserves credit for even addressing this issue.

mgh
December 1st, 2008 | LINK

I just wanted to comment on the bias point, and the economic model.

I think a good book that explains why this isn’t in fact making claims about the malleability of sexual orientation is Richard Posner’s “Sex and Reason.” Posner is a well-respected judge — not a liberal, but not a conservative, really, and a pioneer in law & economics.

In that book, he, too, makes assumptions about how, given a certain set of criteria, more people will choose to engage in certain sexual activities — in essence, one can incentivize or de-incentivize same-sex or different-sexual activity. He’s not saying that sexual orientation per se is malleable, but that human beings make choices based on the perceived costs and benefits of those choices, and in a society in which the costs are higher, they may act differently.

So they’re not saying, more people will be gay; they’re saying, more people will choose to act upon their sexual orientation, and that there’s an increased chance that someone without a preference will choose to engage in heterosexual sexual behavior, as opposed to homosexual sexual behavior.

I know that sounds clinical and somewhat heartless, and it is, sort of, but that’s how econ and science is sometimes, and it’s generally better for it. Posner, as I recall, recognizes that he thinks sexual orientation isn’t malleable, because he says that people choose to engage in same-sex sexual activity despite the enormous costs that it poses to them in some societies. There’s a compassion there, if you look for it.

So, long story short: they’re talking about whether or not people engage in behavior, and not if they’re “turned gay” or not.

mgh
December 1st, 2008 | LINK

I just wanted to add that a good way to think about the above is to consider a society in which homosexuality is punishable by death (cough, Egypt). Clearly, then, even just among gay men, we’d see more of them choosing to live in the closet — that’s an example of (dis)incentives at work. It’s why we work toward a society in which there is no stigma attached to homosexuality — so that people don’t have to deny their sexual orientations.

Ephilei
December 2nd, 2008 | LINK

@Tim
I often filter out reports using general logic, not being an expert, but I’m going to disagree that this should be thrown out. As far as bias, it makes me suspicious but isn’t criteria for dismissal. Lots of good research comes from biased researchers. Arguably, the best scientists are biased because they’ll look hardest. Unless it’s Paul Cameron style irrational bias, it’s fine. We wouldn’t argue with someone with a pro-gay bias.

Regarding the gay heterosexuals, whenever I read something completely mystifying, I give the benefit of the doubt until I can understand it. The situation feels ripe for miscommunication. If it turns out utterly contradictory and critical to the results, then you’re right. But I wouldn’t speak until that happens.

Pender
December 2nd, 2008 | LINK

I think social mechanisms DO play a role — not in whether someone is gay, but in whether he is willing to acknowledge that he is gay and willing to have sex with other men (or acknowledge having done so to researchers).

I definitely believe that gayness is 100% intrinsic, at least to men — but researchers can’t measure gayness directly. They have to rely on proxies, like self-identification and behavior. And those proxies definitely do respond to social mechanisms. I find it completely believable that a disproportionate number of gay men either stopped having sex with men or never started once the AIDS crisis was underway.

Unless there are signs that this guy is not a credible researcher, then if his stuff is getting published in peer-reviewed journals, let’s give it the benefit of the doubt. No need to get anti-science when science says something we didn’t expect or even don’t like.

Timothy Kincaid
December 2nd, 2008 | LINK

quo,

You’re reading what you want to see, not what is here. Francis’ “findings” also show no increase in incidences of homosexuality based on single motherhood. In fact, they showed just the opposite, that single fatherhood was an indicator but that single mothers not at all.

And no, I very much doubt that homosexuality is more prevalent among African Americans.

Timothy Kincaid
December 2nd, 2008 | LINK

Ahem, Gentlemen.

We are not discussing whether Francis’ other papers have merit (i.e. whether or not AIDS or other conditioning factors can have an impact on sexual behavior). I’ve not reviewed those papers thoroughly and have no opinion about them.

If you wish to discuss whether this paper has merit, please do so.

And sorry, guys. But if experience has taught me anything, its that you don’t give the benefit of the doubt when it comes to studies about the etiology of sexual orientation. Most are very flawed – often through no fault of those working in the field.

At best we have a few decent (though sometimes contradictory) studies which provide some tentative parameters for our understanding.

TJ McFisty
December 2nd, 2008 | LINK

“And no, I very much doubt that homosexuality is more prevalent among African Americans.”

Not trying to be funny, but I’ve got a nagging feeling in the back of my mind that Francis just went through and counted all the (B)M4(B)M Casual Encounters on Craig’s List.

I mean, you could get a sampling from that source–doesn’t mean you should.

quo III
December 2nd, 2008 | LINK

“And no, I very much doubt that homosexuality is more prevalent among African Americans.”

Timothy, what are you basing that on? Is it just a hunch, or are there valid statistics?

Timothy Kincaid
December 2nd, 2008 | LINK

Warren,

Thanks for the clarification about confidence intervals. And I’ll agree with you that the variables he’s identified do not seem to be predictive in the real world.

However, I’m not convinced that his work is adequate to cast doubt on other predictive variables as reported or hypothesized by others.

I didn’t go into it in the body of this posting, but I’m just not sure that this sample is adequate to look at the question about fraternal birth order.

The ADDHealth sample consisted of children of which 61% were oldest or single children (the average family with children in 2000 was 1.86).
Only 2.6% of the entire sample (250 kids – male and female) had two older brothers. Of the group only 25 children had more than two older brothers and only one child had more than three.

While that may be representative of the population, it may be too small to tell us anything at all about the youngest of several sons.

Assuming that there is no contribution by a fraternal birth order, we would expect the following results:

116 male children with two older brother would yield 3 gay boys.

12 male children with three older brothers would yeild what, a third of a kid?

If you have two boys who either misidentify the number of male children of their mother or who hesitate to identify as “not heterosexual” and the results are suddenly wildly suggestive in the opposite direction.

Further, Francis’ analysis does not seem to me to adaquately distinguish between maternal siblings and family structure (i.e. half brothers, adopted brothers, and step brothers).

In short, I don’t know that there is much of value at all to be gleaned.

It would have been interesting had there been material and significant findings. But the absense of finding isn’t – as I see it – the evidence that there are no variables at play.

It’s a bit as though Francis went to the garden without a trowel and then reported that he found no potatoes.

CLS
December 2nd, 2008 | LINK

Your concerns seem vague. You discuss “bias” and “motivations” but only provide his conclusions. Are you assuming that conclusions inconsistent with your own indicate bias? Normally bias refers to unproved premises that one basis one’s conclusion upon. For instance, your own premises may be an unspoken bias.

I have not read the author in question but your description does not come near proving your case. You are upset he mentions a “shift from homosexual to heterosexual behavior” in “some men”. Sexual orientation and sexual behavior are not the same thing. There is no doubt that social incentives or disincentives (more typically) shift the sexual behaviours of “some men”. That doesn’t shift sexual orientation only actions — such as so-called “ex-gays” are more likely to stop having sex but no more likely to change orientation.

In addition there are individuals who have sexual orientations that are broader than others. Men tend to fall at the extremes in the Kinsey scale — more so than women. But some fall closer to the middle and thus have no problem “shifting” behavior.

You are most upset by “the assumption that sexual behavior is malleable”. Well, it is. Sexual orientation is a different matter. Are you confusing the two? Now if Francis is saying orientation is changeable that is different but you imply he has not actually said it — only that you think he believes it.

Then to muddy things up further the discussion goes on to sexual identity which is not the same as sexual orientation or sexual behaviour. Identity is how one defines one’s self and that appraisal can be wrong. Individuals who are exclusively attracted to the same sex and only have gay sex could claim a self identity of heterosexual (some exgays do just that).

As for class issues and sexual behavior there have always been huge differences in how people in the different classes (depending on you define them) view sexuality and how they behave. Working class individuals are more conservative in their values but more sexual active. These sorts of differences were noted from the beginning of sexology. And there is no reason to say these studies are a view “best left to porn”. That is just a silly statement.

Again since I have not seen the paper I can’t comment on what the author actually said only on what you say he said. I do know of people who identify as gay but aren’t attracted to the same sex. I remember one such young man who spent hours discussing that very problem with me. He wanted to be gay, saw himself as gay, but was only attracted to women. Whether there are significant numbers of such people I don’t know. Whether such people exist is not questionable.

All my own studies at university and from general life observations indicate that sexual orientation is pretty much determined, sexual behavior can change depending on circumstances (lots of straight me aren’t so straight in prison) without changing sexual orientation, and sexual self-identity may not correspond with the facts (i.e. sometimes people are motivated to believe things that are false). The other thing I know is that there are exceptions to almost all of this since humans tend not to fit into preconceived boxes well.

I tend to agree with “mgh” regarding Richard Posner’s excellent work Sex & Reason.

What I find amazing is that anyone would suggest that people don’t respond to incentives in some meaningful way. If that were the case then buying habit wouldn’t change according price and production levels wouldn’t be influenced by profits. Incentives matter. But that doesn’t mean that rigging incentives (ie central planning) works well as there are always unintended consequences.

Timothy Kincaid
December 2nd, 2008 | LINK

CLS,

In the time it took you to write your rather lengthy objection to my opinion, you could have read Francis’ papers – all of them. Or perhaps you could have taken the time and re-read what I actually wrote rather than respond to what you imagined that I may possibly have written.

What I find amazing is that anyone would suggest that people don’t respond to incentives in some meaningful way.

Yeah, me too. Which is why I didn’t suggest that and why this post isn’t about that subject.

My call for caution (not “my case”) was based on the author’s possible bias, his inconsistent “results”, and his over-reaching conclusions.

Drowssap
December 2nd, 2008 | LINK

The FBO has one thing going for it that no other popular hypothesis has. It’s an evolutionarily sound theory. It’s certainly possible that a small percentage of men might become gay because of their mother’s immune response.

However it surprises me how many well thought out concepts with plenty of supporting evidence ultimately fall apart over time.

I don’t know the paticulars of Francis’ study or his motivations but his large sample size adds some credibility to the studies findings.

If a few more supersized studies have similar findings the FBO will be a gonner. I guess it’s wait and see time.

LegolasTN
December 2nd, 2008 | LINK

While I don’t agree with all of your concerns, Timothy, glancing over the article I do have some major concerns of my own, not the least of which is why he used linear regression to model dichotomous outcomes. According to my statistical training, he should be using logistic regression. The fact that the reviewers of this article didn’t catch this glaring mistake is quite worrisome.

mike/
December 2nd, 2008 | LINK

ah, the beauty of statistics.

i can make them say anything you want them to say. i can even make them say both sides at the same time.

to that, add all of the explanations above about the science and reliability of the study and it makes it even more fun…

Swampfox
December 2nd, 2008 | LINK

Good grief, yet another study. Can’t they just everyone understand that I am gay and I didn’t choose to be gay?

William
December 3rd, 2008 | LINK

TIMOTHY: And no, I very much doubt that homosexuality is more prevalent among African Americans.

QUO III: Timothy, what are you basing that on? Is it just a hunch, or are there valid statistics?

What I understand Timothy to be saying is that there is no sufficient reason to believe that homosexuality is more prevalent among African Americans.

If anyone wishes to maintain that homosexuality IS more prevalent among African Americans, then the burden of proof lies with him or her, not with Timothy.

Chris P
December 3rd, 2008 | LINK

Why is this guy using linear regression on a binary variable? Even with a large sample, I would think that a logistic(generalized linear model) with a binomial link function would be more appropriate. Were interaction effects measured and found to be significant?

There is no discussion on the statistical modeling method that he uses–and by leaving it out this makes statisticians want to discount his entire analysis…

Leave A Comment

All comments reflect the opinions of commenters only. They are not necessarily those of anyone associated with Box Turtle Bulletin. Comments are subject to our Comments Policy.

(Required)
(Required, never shared)

PLEASE NOTE: All comments are subject to our Comments Policy.