Members of the medical community have good reason to be very skeptical of studies with this kind of design. Here's why: It is extremely difficult to exclude the possibility that mothers who breastfeed differ from mothers who don't in very important ways and the boost in IQ is a result of those differences.

Ms. Bazelon states: "Caspi and Moffitt, however, controlled for the confounding factors of social class and maternal IQ", as if that settles the issue.

Firstly, "controlling" for these factors involves some statistical techniques which are of doubtful validity. For example, the most common way to control for unbalanced factors in two groups is to use a multivariate linear regression to tease out the effect of individual factors, which assumes that all the factors (maternal IQ, socioeconomic status) impact on the final outcome (in this case IQ) linearly. If the factor in question does not impact on IQ linearly, then using this technique to "control" for that factor is hogwash. It's done all the time and it's well-accepted scientifically, but it's on much shakier theoretical ground than even most scientists realize.

Secondly, there is always the possibility (in my opinion, a very strong possibility) that mothers who breastfeed differ from mothers who don't in very substantial ways that no attempt has been made to control for. Maybe mothers who breastfeed spend more time with their babies in general, talk to them more, smoke less, adhere to their doctor's advice more - the list of possibilities is endless.

If the above doesn't convince you, consider this. It was precisely studies of this kind that purported to show that hormone replacement therapy for post-menopausal women saved lives. Doctors were reassured that every conceivable confounding factor had been taken into account and "controlled for", with the final result being that hormone replacement was a great thing. Women who got hormone replacement in these studies DID live longer and get fewer diseases.

But when more rigorous studies were done, it turned out to be completely wrong. Hormone replacement therapy didn't save lives - it killed! The reason women who took it lived longer in those earlier studies was because they were DIFFERENT than the women who didn't in many other ways that had nothing to do with hormone replacement therapy, despite all the researchers' assurances that everything had been "controlled for".

As a result, a generation of women was exposed to needless risk based on faulty science, and thousands probably died. In my opinion, it is one of the great shames of modern medicine. It was exactly the same kind of study design that led to that tragic result, so think twice before believing stuff like this.