1. Small sample sizes.
2. Significance was only found in one subset of the population. Given enough ways to divide up groups of people, you will almost always get false positives.
3. Not an experimental design.
4. The study did not control for simple factors like income or the health of the parent.
That is not fair. You're making it sound like it's a JS framework.
Seven years is nothing in this case.
A single data point isn't science. It -might- be news, if it was a super recent data point (though, obviously, requires peer review and replication). This is neither.
I'm not familiar with the research here so I can't say either way. But a preliminary google scholar search turned up a lot of articles which seem to suggest that breastfeeding has a bunch of positive impacts on numerous intelligence and general well-being metrics.
The article itself has been cited 209 times according to google scholar [1], and this study [2] (cited 400+ some odd times) seems to accept some of the propositions made in this paper as givens.
Anyway, I think it would have been interesting if the null hypotheses here weren't rejected. I kind of expect evolution to have optimized human breastmilk over alternatives for infant development.
1. https://scholar.google.com/scholar?cites=1509419575151008169... 2. http://pediatrics.aappublications.org/content/129/3/e827.sho...
I realize that repeat studies are important generally in science. But, many studies (particularly health related) seem to have such a result built into their design. If you get a result, it will just be preliminary and require a larger sample study moving on.
Is there a mechanism in science for actually taking these preliminary results and validating/falsifying them more conclusively? Does a result like this make a larger scale study this team's top priority for the next study? Will these results convince another team to do the full study?
Otherwise, it feels like there's something structurally wrong. If the study is designed as a non-conclusive prototype or framework, what's the point unless it leads to other studies?
I'm also not sure what you mean by #3.
Since the mothers self-selected breastfeeding, the study makes the assumption that banked breast milk (from a random donor) is equivalent to fresh milk from the mother.
It assumes that the act of suckling on the mother's breast has no bearing on brain development. See recent evidence regard pre-term babies and skin contact: http://handtohold.org/resources/helpful-articles/the-benefit...
Finally, it ignores other (more recent) studies showing how the mother's milk reacts to signals from the baby's saliva (backwash from baby sucked back into breast). https://www.sciencenews.org/blog/growth-curve/backwash-nursi...
(Not saying these links aren't equally or more flawed, but it's avenues of causation ignored.)
Wouldn't a valid study require randomized instruction to mothers about whether to breastfeed? Is that ethical? Compliance is equally tough: breastfeeding can be incredibly difficult for some moms/babies. Further, couldn't the difficulty in breastfeeding also be correlated to the effects -- could children reject breastmilk based on the mother's health or otherwise based on the "quality" of the milk?
Please forgive if I'm asking questions that are well understood -- by no means am I claiming to be knowledgeable in this field. But I do have kids and went through hell with one of them on this, so I'm very interested in the research.
It would be much more useful to talk about the statistical significance in more quantitative terms.
Because it's my understanding that the evidence shows a breastmilk->IQ correlation as well as all the other health benefits [https://en.wikipedia.org/wiki/Breastfeeding#Baby]