.comment-link {margin-left:.6em;}

Monday, July 24, 2006

 

Bearman and Bruckner: There is no fraternal birth order effect

Peter Bearman and Hannah Bruckner authored a paper in 2001 that was later published in the American Journal of Sociology, vol. 107, 1179-1205, 2002 which offers a major challenge to the recent fraternal birth order effect suggested by Anthony Bogaert. Bogaert's research, published in the Proceedings of the National Academy of Science of the USA, suggests that having biological brothers, not non-biological ones (adopted, step-brothers) is the key to this effect, thus suggesting pre-natal factors, as yet undiscovered.

Those interested in this line of research should examine Bearman and Bruckner's study which I believe to be very well done. They were also able to catch identical twin data and found low concordances for MZ, DZ and siblings. There were no significant differences between groups.

More on this paper later...

Comments:
I presume your announcement will be somthing along the line of Bearman and Bruckner confirm Kinsey's "10% of men are homosexual"

No? Oh well, never mind.

And what? Is this subject area more interesting than the Bearman and Bruckner research -- same authors, same dataset -- that showed abstinence pleges etc were a menace to the health of youths? :)

(Keep in mind I've been chopped off from access to the entire dataset... and the "free" data doesn't include the sexuality responses... so don't ask me to go retrieve any figures. But here goes from memory.)

The questions in the dataset about sexuality are unstable across time. And definitionally a bit squishy. These were rather broad questions answered by teenagers, afterall: an introduction of confusion/denial/inaccuracy etc you've had no hesitation accepting in the past. Most twin-studies interview adults who have settled and more accurate patterns of both attractions and labeling of those attractions.

I don't think the low absolute numbers would support any meaningful comparisons if we start snipping it up into into sub-groups, but it would be interesting if the B&B outcomes were also unstable as a direct result.

It is also impossible to obtain measures of, say, K-scores from the dataset. At best one can assume "only same-sex" or "same and opposite sex" attractions, and it appears B&B have lumped both together. Not sure that's a good idea, particularly given the unstability of the responses.

The results in other areas of the dataset are very interesting, but do also highlight the difficulties when interviewing a population still searching for meaning and definition. While showing up the possible concerns that young people have, the figures across questions such as "will you be killed by age 21" or "will you get HIV/AIDs" bear little relation to the actual risks or outcomes.

So, you have our permission to tell your kids that "research proves that young people have distorted ways of seeing things" the next time they disagree with one of your decisions.

Ha ha. And I bet they'd calmly agree and acquiesce too.
 
I presume your announcement will be somthing along the line of Bearman and Bruckner confirm Kinsey's "10% of men are homosexual"

No? Oh well, never mind.

And what? Is this subject area more interesting than the Bearman and Bruckner research -- same authors, same dataset -- that showed abstinence pleges etc were a menace to the health of youths? :)

(Keep in mind I've been chopped off from access to the entire dataset... and the "free" data doesn't include the sexuality responses... so don't ask me to go retrieve any figures. But here goes from memory.)

The questions in the dataset about sexuality are unstable across time. And definitionally a bit squishy. These were rather broad questions answered by teenagers, afterall: an introduction of confusion/denial/inaccuracy etc you've had no hesitation accepting in the past. Most twin-studies interview adults who have settled and more accurate patterns of both attractions and labeling of those attractions.

I don't think the low absolute numbers would support any meaningful comparisons if we start snipping it up into into sub-groups, but it would be interesting if the B&B outcomes were also unstable as a direct result.

It is also impossible to obtain measures of, say, K-scores from the dataset. At best one can assume "only same-sex" or "same and opposite sex" attractions, and it appears B&B have lumped both together. Not sure that's a good idea, particularly given the unstability of the responses.

The results in other areas of the dataset are very interesting, but do also highlight the difficulties when interviewing a population still searching for meaning and definition. While showing up the possible concerns that young people have, the figures across questions such as "will you be killed by age 21" or "will you get HIV/AIDs" bear little relation to the actual risks or outcomes.

So, you have our permission to tell your kids that "research proves that young people have distorted ways of seeing things" the next time they disagree with one of your decisions.

Ha ha. And I bet they'd calmly agree and acquiesce too.
 
Sorry, that will read confusingly.

The "unstable across time" refers results from the different age groups, their non-response rates etc.

Without knowing the response of any individual there's no way of knowing how many individuals are stable across, say, the three waves of surveys. Some no doubt are. Some not.
 
These authors are equal opportunity challengers. The ADD Health data base is about as good as we have. Any study can and should be challenged but this is a representative sample versus the convenience samples of Bogaert and Blanchards. The one population based sample of Bogaert showed a marginal result and an effect size (explanation of variance) of .25%. The more representative the sampling the less the FBO shows up. This is not to say that I think Bogaert deliberately uses convenience samples to skew results. I think he is a curious researcher who thinks he is on to a small piece of the total puzzle. He has been very helpful in our correspondence about his work.
 
Just noticed Warren -- there's two of my post.

This is one of the few times we don't have to say "Sorry about the double post folks"...

... because you did it! :)

------------------------------

And are you saying... you are OK with the methodology of using the indirect questions asked of younger teens etc as an accurate measure of sexual orientation?

(ie the stable adult S.O.?)

Given you've claimed elsewhere that even a direct "I am 100% gay" response by a young person is not something that should be believed/accepted without reservation ... the guidelines as one place for example ... this does seem rather curious of you.
 
No, of course those who report SSA would be the group to generalize to. However, I think most people would recognize SSA first and then through some behavioral path eventually end up at a gay identity. Bearman and Bruckner do note that their study does rule out the possibility that there might be genetic/hormonal factors that provide the push toward behavior and later identification. I do not disagree with this. We have been calculating predictions of variance from the various biological factors that might predict adult SO and have found nothing that predicts more than 8% of the variance (X chromosome skewing of women with two sons predicts 8% of why their sons SO). The birth order effect only predicts about .25 - 5%, depending on the sample and question asked. Gender nonconformity predicts 15-40% again depending on the sample.
 
As I'm the oldest child, only son and did not display gender non-conformity you'll have to keep looking. Obviously :)

[And that pathway is like soooo yesterday. More common now -- if we go on ages -- is attraction(12)->identity(14-15)->behaviour(16-17). Basically mirroring their straight peers, and why the age of coming out has fallen in recent years even though the other ages have not. I think this is in Savin-Williams.]

Now... you're not about to embarrass yourself by doing only unifactorial calcs I hope.... ???

"What causes one to feel cold?"

"Well after extensive literature search we found that the wind speed only accounts for 5-15% of feeling cold. The weight of your jumper can only account for <2%. So we postulate that clothes are an unrelated factor, or a minor contribution at best. Living in Canada could only cause 20-40%."

"What about the air temperature?"

"The air... what???"
 
The X chromosome guys don't seem embarrassed, nor did Bogaert. Of course, the whole thing is multifactoral. This is actually my point.
 
"The birth order effect only predicts about .25 - 5%, depending on the sample and question asked."

OK, I'm going by memory here, but I'm pretty sure you said earlier that it predict 1/7 and I think that I read that the this was an early projection and that closer evidence said 28%. I'm thinking that came from "Born Gay" but I may have my source incorrect.

Timothy
 
Those are separate issues. The estimate of numbers of men that may owe sexual orientation to birth order is an estimate from Blanchard and cannot be taken as gospel since it is an estimate.

The effect size of the studies are statistical estimates of the variance accounted for by the relationship between the variables. These effect sizes are quite small but are important in that they allow us to look at one of what are probably many factors that may predict sexual orientation. I probably should not sound so certain about these effect sizes as yet. I have been talking to Anthony Bogaert about this some and he agrees that the effect sizes are quite small. He did not calculate an effect size for the recent study but my colleague Gary Welton did and he came up with 1%. We are documenting all of this and will run it by Bogaert and others as well as peer review before we say much more.
 
Sorry. Thought you meant occurance of effect, not effect size.

When you've had the effect size peer reviewed, please let us know the method used as my (VERY limited) understanding of effect size seems to recall that the same numerical result can be either very low or very high depending on the method.

Timothy
 
Post a Comment

Links to this post:

Create a Link



<< Home

This page is powered by Blogger. Isn't yours?