Did Chinese Children Evolve To Take Tests After Breakfast?

My recent posts have focussed on IQ and the differences between a gap in IQ test results and a gap in general intelligence or g. The contemporary difference between white and Black racial mean IQ is about 10 points. For every IQ test subject, including all white subjects and all Black subjects, some portion of that IQ score represents a measurement of g. However, there are good reasons to think that the proportion of the IQ score that measures g will be different among white people from the proportion measuring g among Black people. While I don’t think that motivation is different enough between racial groups to explain the mean IQ score gap, it’s very interesting and relevant to note that placebo effects that are likely due entirely or almost entirely to motivation effects (primarily a combination of arousal effects and attention effects) can on their own generate a 10 point difference in mean IQ test scores. The Proceedings of the National Academy of Sciences (in this case, the “nation” is the USA) has the lowdown.

10 points is a pretty big deal. We’re talking 2/3rds of a standard deviation. This work suggests that holding a pep rally before standardized testing could easily make a school seem much better than its actual academic accomplishments would merit. If you run a charter school, there’s good reason to pay attention to the result.

But it doesn’t give us anything that is immediately useful for making policy that improves the lives of people generally. In that sense, it’s a lot like the work of Charles Murray. Much of what Murray has to say is accurate, but in addition to using that accurate information as premises is fallaciously reasoned arguments that lead to erroneous conclusions, there simply isn’t much there that is policy-useful. Not in his true premises, and not even in his erroneous conclusions. (In fact, Murray often goes out of his way to insist that he’s not recommending policies, but merely reporting scientific information.)

That doesn’t mean, however, that IQ testing research hasn’t generated policy-useful information. We already know that, for instance, lead from paint, solder, and industrial uses needs to be cleaned. We know this, in part, because children have had their lead exposures measured and then been tested using a variety of cognition and problem-solving tests. We’re confident lead is the problem because of the consistent finding of a strong negative correlation between lead exposure and IQ scores which we wouldn’t have without IQ research combined with the neurological research showing lead’s toxic effects and ultimately demonstrating a clear causal pathway. The information we have could easily justify a much more intense public effort to complete that work than we’ve seen to date.

But there’s more useful information than that. Today I want to bring your attention to a study of children in China. Partially funded by the NIH (again, a USA institution) and bringing together work by professional researchers in the USA and China, the study looked at whether research in the US showing a positive correlation between regular breakfast consumption and IQ scores was generalizable to other populations that are genetically and culturally distinct. It also attempted to put the correlation on firmer footing by using tests of cognition and problem solving that were much more broad than some of the narrow or even skill-specific testing used in early research on the importance of breakfast to young students in the USA.

In other words, is breakfast only important to people in the USA? And does it only affect specific skills, or does it have an actual impact on g?

The results were clear: for kindergarten-age Chinese students, eating breakfast almost every day was strongly (p < 0.01) and positively correlated with an increase in scores on a general intelligence test when compared to same-age Chinese students from the same city who ate breakfast only “sometimes”. This was true

even after adjusting for gender, current living location, parental education, parental occupation, and primary child caregiver.

The naked effect size is summarized by their Table 2:

Liu et al. find a difference in Verbal IQ between Chinese children who eat breakfast "sometimes" and those children who eat breakfast "often or always" of 7.9 points. The difference for Full IQ is 7.8 points.

This table taken from “Regular breakfast consumption is associated with increased IQ in kindergarten children” by Liu, et al. Downloaded from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3606659/

However, this still needs to be controlled for factors mentioned above as known possible confounds. Table 3 lists the robust results that remain after multivariate analysis adjust for the effects of variables such as gender and parental education:

Liu et al. find a robust difference in Verbal IQ between Chinese children who eat breakfast "sometimes" and those children who eat breakfast "often or always", even after mathematically controlling for other variables. The difference in Verbal IQ after analysis was found to be 5.6 points. The difference for Full IQ is 4.6 points.

Table 3, taken from “Regular breakfast consumption is associated with increased IQ in kindergarten children” by Liu, et al. Downloaded from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3606659/

You can see that the scores for PerformanceIQ (or non-verbal/other-than-verbal IQ) shows a gap, but that the p-value tells us this is not particularly robust. Nonetheless, the p-values for both Verbal IQ and Full IQ are less than 0.01. This is a good, solid result for such studies, and is probably the result of getting a good, appropriate sample of solid size and other sound methodological choices. The study, especially as it is itself an extension of similar studies in the USA, should be easily replicable and we have no reason here not to treat the results as reliable.

While the discussion of the paper is conservative about their ability to determine the mechanism by which breakfast provides these benefits, the benefits themselves really aren’t in doubt. Some theories of causation can also be ruled out. Contrary to evo-psych devotees or Murray-like thinkers, the study does not show a deep genetically-determined preference for taking tests right after scarfing some food or tell us anything about a racially distinct population-genetic effect on intelligence tests (in fact, part of the reason for doing the research in Jintan, China was to get a largely mono-racial sample).

What this paper tells us is that our kids need food, badly. While other IQ research is bent and twisted (or disdained and dismissed) for a wide variety of reasons, this and similar results give us an additional scientific justification for what so many people already believe we should do: feed the hungry, dammit.

I’ve covered on this blog the (now defunct) policy in Seattle Public Schools to publicly shame children whose parents were behind on lunch payments. While we ended the policy of shaming children for their parents’ poverty (and/or negligence, I guess it might be in a tiny portion of cases), there is still a strong resistance to the idea that we should feed all children at our schools as a matter of course. Even where strong lunch programs exist, breakfast programs are frequently dramatically underfunded or entirely absent. Moreover, most school breakfast programs charge for breakfast, and even states that are doing a relatively good job (such as Florida, surprisingly) may have laws that only mandate schools make available a breakfast program for elementary schools, not middle schools or high schools. Schools for older students aren’t prohibited from offering breakfast, but may not do so since they aren’t required to and because the revenue generated from charging students for breakfast isn’t generally enough to make the program revenue-neutral.

But it’s time to stop worrying about making these programs revenue neutral, and start offering free, nutritious breakfasts for all students. Yes, many students may not need it. Yes, many students may have food available at home that they prefer to eat (since they frequently have more influence on what parents buy from the grocery store than on what schools serve in their cafeterias). But a free, non-stigmatized, come-as-you-are, all-you-can-eat breakfast would improve long term learning as well as provide nutritional and other effect that might just improve that fundamental general intelligence with which Murray and others are so obsessed.

Also, my folk-punk band is available for hire to any school district that wants to add a cool, counter-culture cachet that will get kids in the door of cafeterias that otherwise might not be seen as a hip place to hang.

Just trying to do my part.


  1. Maya says

    I find it distressing that we aren’t doing common sense actions as a society like making sure that at least children can get enough food to eat already.

    We really shouldn’t need statistical research to know that people (especially children) do better when their basic needs are met, as I would think this is would be a basic fact of human existence for most adults.

    Mentioned at the end of the paper is the possibility that sleep quality might be a confounding variable. Sleep is another basic need that we seem to neglect far too often as a society.

  2. Crip Dyke, Right Reverend Feminist FuckToy of Death & Her Handmaiden says

    Fully and completely agreed.

Leave a Reply

Your email address will not be published. Required fields are marked *