Passed on to me via email from a friend who reads Slate who wrote the researcher. Try again Emily B.

To Researcher:

Slate has taken your data and said your conclusion(s) are fairly weak in part because of the low number of third born sons (81) and not taking into account birth weight. Is their interpretation of the data accurate?

From:

Thanks for the response.
The number of third born with two elder sibs who died is low, which is apparent from the wide confidence limits (Figure 1 in Science article). This is nevertheless highly significant, so it is a matter on how you define "fairly weak". I would rather say that the adjusted result in this small group is in agreement with the conclusion, but it is not robust.
We did take birth weight into account (see Science figure 1 legend and the SOM material).

Sincerely
Petter Kristensen

I don't know anything about the study, etc, but the second part of the question seems to have missed Emily's point. The problem was that the study did control for birthweight in the regression analysis.

In a multiple regression analysis, you estimate the effect of a one unit change of one variable (the independent variable, in this case birth order) on another (the dependent, IQ), while keeping other variables (controls, including birthweight) constant.

The problem is, what if the effect birth order has on IQ is mediated through a biological process which has birthweight implications? By controlling for birthweight, you are essentially precluding that possibility and the measured effect that birth order will have on IQ will thus not reflect those types of processes. The model specification in fact implicitly assumes there is not a biological factor at work, which, given the research question, should have been left open.

I don't have the research either. But the issue seems to have been about TWO older siblings dying in order to study the 3rd and not one in which only a single older sibling died. I'm assuming it wasn't a typo and that he matched up the "81" number with the relevant data.

Also, note the IQ of the second sibling after the first dies went up commensurately as if he was first born (only from reading news article online). I'm not sure how that was determined. Did it include birthweight somehow?

But thanks for the clarification on controlling for birth weight in a statistical sense :-)

This excellent response is from Judith Rich Harris:

Thanks, jasamcarl. Yes, that's exactly what I meant: that the
researchers should NOT have controlled for birthweight. You
explained it beautifully.

In the email I sent to Emily before she posted her piece
("Sibling rank and IQ") on the XX Factor, I mentioned several
reasons why I favor a biological explanation for the IQ difference found in the first Norwegian study, and why I wasn't impressed with the second Norwegian study - the one that allegedly showed that the birth order effect was social, rather than biological. Emily's piece mentioned some of my reasons, but not all, and not necessarily the strongest ones.

Let me give you three other reasons why I think the IQ difference - the slightly lower IQs found in boys with older brothers - is biological. First, there is other evidence that the birth of a boy has deleterious effects on subsequent children born to the same mother, and that these effects are biological in origin (due to being gestated after an older brother, not to being reared with an older brother). The link that Emily supplied at the bottom of her piece connects to an article in _Scientific American_ that describes some of these effects. For example, children born to mothers who previously gave birth to a boy are physically smaller
and have a greater chance of dying from an infectious disease.

Second, one of the results found in the first Norwegian study
(but not mentioned in the press coverage) fits poorly with the
usual explanations given for the birth order effect. The study
showed that the IQ difference between the firstborn and
secondborn was WIDEST when they were born LESS THAN A YEAR APART. This finding doesn't support the theory that firstborns are smarter because they receive more parental attention: How much attention does a toddler get if his parents have a newborn on their hands? It also doesn't support the theory that firstborns are smarter because they act as tutors for their younger siblings. Surely a firstborn two or three years older than his sibling is in a better position to serve as tutor than one who is only 11 or 12 months older. But the IQ advantage for firstborns who were two or three years older than the next sibling was smaller than that for firstborns who were 11 or 12 months older.

On the other hand, the finding that secondborns do worse if they are born shortly after a male firstborn is quite consistent
with a biological explanation. Whatever toll the birth of a boy
exacts on the mother, she tends to recover from it over time.

Finally, I was not convinced by the data analysis carried out
in the second Norwegian study, the one reported in _Science_. Dr. Kristensen said that his results were highly significant (in the statistical sense) and I'm sure they were. However, the data on which the tests of statistical significance were performed were the "adjusted" IQ scores - "adjusted for parental educational level, maternal age at birth, sibship size, birth weight, and year of conscription." Making such adjustments is a legitimate part of statistical analysis but it involves some risks. We've already mentioned one: it might not have been appropriate to "adjust" for birthweight. The researchers admitted in their paper that the reported
associations between birth order, social order, and IQ were
"stronger in adjusted models." In other words, the results
they were reporting would not have looked so convincing if
they hadn't made these adjustments in the data.

This study has not yet been replicated by other researchers
using different groups of subjects. Neither has the first
study - the one that reported a 3-point difference in IQ
between firstborn and secondborn boys. I think we should
wait for replications before we jump to any conclusions
about the intelligence of firstborns and laterborns. In
psychology, as in medicine, interesting early results often
fail to hold up over time.

Harris said: "[The finding that IQ difference was widest between brothers born less than a year apart] doesn't support the theory that firstborns are smarter because they receive more parental attention: How much attention does a toddler get if his parents have a newborn on their hands?"

Not necessarily true. First, the obvious explanation still holds: it may be that the undivided parental attention a first-born receives during that FIRST YEAR where there is no rival is quite important in developing the building blocks for IQ.

Second, I would suggest that an older brother who is 1-2 years older demands a lot more parental attention than an older brother 3-4 years senior. The former is dangerously mobile (toppling down stairs and licking wall outlets come to mind) and hasn't developed the ability for much self-directed play, while the latter is much more independent, has a longer attention span, and requires less direct interaction from the parents. Thus, a younger brother competing with a sibling close in age will get less parental attention.

On the whole, I believe the evidence still favors an environmental explanation.

Cool name and nice points Stan

I'm sure this is mind-bogglingly complicated and undoubtedly multifactorial. I can't imagine conceiving of a study that would actually answer this question and hey...I'm first born! Don't know about the researchers...maybe they should get their older brothers to look into it.

Also, note the IQ of the second sibling after the first dies went up commensurately as if he was first born (only from reading news article online). I'm not sure how that was determined. Did it include birthweight somehow?

Did your expert account for this or did I misunderstand the reporting of this siblling birth order study.