The Washington PostDemocracy Dies in Darkness

Paper finds a surprising link between warmer temperatures and math test scores


We all know about some of the more obvious consequences of warmer weather — like heat waves.

But poorer math performance, at least in the short term when kids take tests on particularly hot days?

That’s a surprising finding from a new National Bureau of Economic Research working paper by economist Joshua Graff Zivin, of the University of California, San Diego, and two colleagues, which the authors call the “first economic analysis of the relationship between temperature/climate and human capital.”

Drawing on existing literature on how warmer temperatures can affect the brain, the researchers proceeded to examine children’s test scores from the National Longitudinal Survey of Youth, a very long-running study that began with children born between 1957 and 1964, and since 1986 has also involved assessments of their kids.

The researchers were able to study 8,003 children who had been given recurring math and reading assessment tests — sometimes on more than four separate occasions over time — in their homes. The dataset allowed them to overlay test scores with the average temperature in the county where they lived on the day of testing. And hence their very surprising result — “we find that math performance declines linearly above 21C (70F), with the effect statistically significant beyond 26C (79F),” as the paper puts it. No effect was found for reading scores, however.

“If you move from roughly 70 degrees Fahrenheit to roughly 87 and a half degrees Fahrenheit, a child’s mathematics score decreases by 1.6 percentile points,” said Graff Zivin. “We throw billions of dollars at moving test scores in the United States, and we do a reasonably lousy job of it…so when you see a test score change of this magnitude, it’s pretty notable.”

To seek out a reason for this effect, the paper turns to neurological literature on the effects of temperature on the brain. The brain uses a great deal of energy and creates heat in doing so — which the body dispels less efficiently when it’s hot out. Thus, the researchers suggest, a variety of mental operations, such as the use of working memory, may be affected by warmer temperatures.

“Military research has shown that soldiers executing complex tasks in hot environments make more errors than soldiers in cooler conditions,” they noted. They also hypothesized that this would affect math scores more than reading scores.

One of Graff Zivin’s co-authors, Solomon Hsiang of the University of California, Berkeley, has also previously co-authored research suggesting that economic productivity declines on hot days – although this is likely more a physical than cognitive effect. Thus, the new research is part of a body of new economic inquiries into the potential toll that rising temperatures may take on humans’ productivity and intellectual capital.

[People don’t work as hard on hot days — or on a warming planet]

But here’s where the study took a surprising turn. The authors also examined whether there would be a cumulative, long-term effect on students’ scores from temperature. They thought the impact would compound, really leading kids’ scores to take a nosedive over time.

But in a long-term analysis of scores over time, they didn’t find that. Rather, they found that temperature didn’t have a significant effect any longer.

“I’ve got to tell you, we puzzled over that for at least a year, to try to really wrap our heads around whether we’re doing it right and what might explain such a pattern of results,” said Graff Zivin.

The researchers ultimately concluded that, much like Florida building its beachside roads higher up in the face of rising seas, parents and children were probably engaging in a kind of “adaptation” to higher temperatures (even though they were not aware that’s what they were doing).

“You get a hot day on the day of your test, you take the test, you discover that you’re several percentiles lower,” says Graff Zivin. “Then your parents say, ‘no more field hockey, you’re going to Mathnasium.’” In other words, a kind of “compensation” kicks in, and families — driven by parents — may strive ever harder to bring test scores up, and counteract the effects of temperature.

“Parent[s] are offsetting more than a 6 percentile point accumulated decrease in human capital due to warmer temperature exposures,” the authors wrote.

This is not, the authors noted, without its cost. It takes considerable parental effort to drive up math scores. “The fact that people are engaging in lots of compensatory behavior, because they’re sending their kids to math camp, that has social costs,” said Graff Zivin.

One possible objection to the study, however, involves air conditioning — most U.S. homes  have it, and the question is whether this could somehow insulate kids from outdoor temperatures when they take the test. But Graff Zivin doesn’t think that can explain away the results.

“We don’t know what the kid is doing 2 hours before the test implementer arrives at the house, it’s quite possible the kids are outside,” he says. “It’s quite possible these kids are exposed to some heat even if they can then come in and cool off.”

The takeaway from the study, in the end, is rather paradoxical. On the one hand, it extends the inquiry of the impacts of a warming planet into a fascinating realm that we rarely think about — the cognitive. And these sorts of effects are unlikely to only show up in children taking math tests. They could happen all across our world and our economies.

At the same time, though, the research also shows that the human capacity for adaptation and adjustment over time can be a powerful thing. We may or may not be able to hold back rising seas, but it seems we can drive up our kids’ math scores.

Also in Energy & Environment:

This ice shelf is nearly the size of Scotland — and scientists are worried about its stability

No, genetically modified pet fish are not going to wreak ecological havoc

Troubling new research says global warming will cut wheat yields