Summer is upon us, which means an increase in street crime and ice cream consumption. However, neither one causes the other; they are both highly correlated with summer’s heat, which brings more people out of their homes and onto the streets, where some eat ice cream and some get mugged. Correlation is not causality.

Here are two more facts to ponder: American children take lots of standardized, machine-scored, multiple-choice tests, and they are getting fatter. Is this just another correlation, or could one be a cause of the other? Could excessive testing be at least partially responsible for the increase in child obesity?

What makes this issue complex are two other variables, an increase in poverty and the disappearance of school recess.  This sad and entirely avoidable situation also illustrates the unfortunate truth of the maxim, “What we don’t care enough to measure does not matter.”

No question that obesity is on the rise. An astonishing 18.5% of American youth ages 12-19 are obese, and 5.6% are severely obese. If we include children who are overweight but not necessarily obese, the situation becomes direr. “31% of children ages 10 to 17 were categorized as overweight or obese. This statistic varies slightly by gender, with boys more frequently affected than girls (33% of boys versus 29% of girls).” That study by the Annie E. Casey Foundation ranks states, ranging from Rhode Island’s 39% to Oregon’s 16%. One more number from that study: Nearly half (47%) of American children do not exercise regularly.

Seven out of 10 overweight adolescents grow up to be overweight or obese adults, and the consequences are grim: high blood pressure, high cholesterol, abnormal glucose tolerance, heart attacks, and diabetes. The latter often means other serious health issues like blindness and early death; diabetes and its complications kill about 200,000 Americans every year.

Poverty and obesity are positively correlated, unfortunately. The poorer a child is, the more likely he/she is to become obese. “For a long time researchers have tracked high rates of obesity among black and Hispanic kids, but a closer look at communities shows family income matters more than race in predicting which kids are overweight.” Based on data about 111,799 Massachusetts students in 68 school districts, a study by the University of Michigan Health System showed that as poverty rises, so does the rate of obesity among children.

Testing is also increasing, making American students the most tested in the world. As Harvard’s Dan Koretz, author of “The Testing Charade: Pretending to Make Schools Better,” told me in an email, “Just the requirements in reading and math under the Every Student Succeeds Act mean being tested 14 times, and that’s the tip of the iceberg in many locations because of all of the interim and benchmark testing.” He added that countries with a reputation for being test-centric, such as Singapore, test students only two or three times during a student’s years in school.

Like obesity, testing correlates positively with poverty: the poorer a child, the more time he/she will spend being tested or practicing test-taking. According to a 2016 survey of teachers, Listen to Us: Teacher Views and Voices, 26 percent of teachers devote more than a month to test prep, and “A greater share of teachers in high- and medium-poverty schools reported spending more than a month on test-prep activities for district and state tests.” That’s at least one-eighth of the school year, and, since it all comes at once, it must seem like an eternity to those low income students and their teachers.

So, we have lots of overweight kids taking lots of standardized tests, but here’s where it gets interesting: Because the length of the school day is fixed, in order to increase testing and test-prep time, schools had to eliminate something. Sometimes the arts and science were slashed, but often the first to go was free play time, a.k.a. recess. The pressure to improve standardized test scores was particularly intense in low income communities, which fixated on ‘the achievement gap.’ Atlanta, for example, eliminated recess entirely. ”We are intent on improving academic performance,” Superintendent Benjamin O. Canada, told The New York Times in 1998. ”You don’t do that by having kids hanging on the monkey bars.”

(Incidentally, the no-recess policy was continued by Canada’s successor, Beverly Hall. During her tenure APS was engulfed in a massive cheating scandal. Here’s an irreverent question: Does lack of recess for children merely correlate positively with cheating by adults? Or is it a cause?)

The disappearance of recess is a plausible explanation for the epidemic of childhood obesity. Of course, there are other culprits, including too much screen time and fast and processed food, but lack of exercise–remember, 47% of children don’t get regular exercise–plays a huge part.

Unlike obesity and testing, recess time is negatively correlated with poverty. Simply put, richer children get more time on the playground. The poorer the children, the less time on the playground. And, sadly, cutting recess does double damage to many inner city kids, whose parents are loath to let their young children play outside after school because of dangers, real and perceived. So no play time at school may mean no daily exercise at all.

A chain of causality seems to be emerging: Excessive testing causes cuts in recess, which then contributes to widespread obesity, and poverty makes everything worse. That’s merely speculative, so let me suggest a fact-based alternative chain: Regular recess leads to better physical and mental health (i.e., no obesity and better academic performance).

And that’s not speculation, causing one to wonder about the mental acuity of educators who did away with recess. How could they not know that regular exercise pays dividends, that it reduces the risk of obesity, provides socializing opportunities, and promotes mental agility and improved academic performance? Don’t the recess-cutters understand that improved physical fitness is positively correlated with better performance on standardized tests and higher grades? As the experts put it, “Studies have found that lifestyle changes and small amounts of weight loss in the range of 5-10% can prevent or delay the development of type 2 diabetes among high-risk adults. Lifestyle interventions including diet and moderate to intense physical activity (such as walking for 150 minutes per week) were used in these studies to produce small amounts of weight loss. The development of diabetes was reduced by 40% to 60% during these studies, which lasted three to six years.”

Another intervention study found that providing overweight children with 40 minutes of physical activity increased cognitive scores. School-based physical activity can improve students’ attention, concentration, and ability to stay on task. And kids who get to run around and burn off energy behave better in class.

The World Health Organization says children need 60 minutes of moderate to vigorous physical activity every day, a goal that regular recess at school could help achieve. Unfortunately, most public schools do not provide recess. Only 11 states require either recess or “general and physical activity” during the school day. New York is not one of them, but the New York State Education Department does have a PE requirements of 120 minutes per week. Physical education must be taught by a certified teacher who provides instruction according to New York State PE standards, but, even if recess is provided, it doesn’t count toward that 120 minutes per week.  In the District of Columbia, the Central Office has a “Wellness Policy” calling for at least one 20-minute recess period per day for ‘child-initiated discretionary time,’ but it’s not clear whether schools actually adhere to this, or if there are consequences for not falling in line.  Students in Texas get on average only 20 minutes of recess per week, a decline of over an hour since 2001, which was the year the test-focused No Child Left Behind became federal law.

We’ve known about the importance of regular exercise for a long time. Presidents from Dwight Eisenhower on promoted physical fitness among young people. Ike established the President’s Council on Youth Fitness in 1956. President John F. Kennedy famously took 50-mile hikes to promote exercise. Lyndon Johnson created a special award for 10-17-year olds, the Presidential Physical Fitness Award, and enlisted baseball superstar Stan Musial to attract attention to the cause. Under Ronald Reagan we got a postage stamp honoring physical fitness. And so on.

These campaigns about the value of exercise had no appreciable impact on schools, which generally don’t respond to exhortations, only to pressure. Our schools report only what we tell them to measure, and society’s consistent message has been: Tell us your test scores!   Those scores, generally speaking, are the only educational measurement with real consequences, and they are used to reward and punish educators, schools, and students.

Assuming that no one wants children to become obese or even overweight, schools must provide regular recess, but that won’t happen unless it’s mandated and measured. Instead of just providing test scores, schools must be required to report the answers to two more consequential questions: “How many hours of recess do students have each week, and in how many separate segments?” We should provide incentives (such as playground repairs where needed) and at the same time make it clear to principals that if they fail to provide recess, they will be penalized. Before long, more children will be out on the playground or in the gym, playing.

But this will not happen until recess is both mandated and measured.

While how much recess children should have, and how many times each day, are local decisions, educators might want to look to other countries for examples: As the USPlay Coalition explains, “Japanese children get 10-20 minute breaks between 45-minute lessons or five-minute breaks and a long lunch. Finnish and Turkish children have 15 minutes to play after each 45 minutes of work. Ugandan students have an eight-hour school day, but they have a half hour of play in the morning, one hour for lunch and play, and 1.5 hours of activity time (sports, music, art, free-choice playtime) in the afternoon.”

Establishing that recess matters–with incentives for success and consequences for not measuring up–sets an interesting precedent. Perhaps we should also ask ourselves what else besides academic performance and recess matters. If we decide that art, music, and drama are important, let’s insist that schools provide them and then take pains to measure the hours of opportunity students have to pursue them.

Measuring academic achievement is clearly important, but the academic health of a school can be determined by testing a well-drawn sample of students. We can test less and still know what we need to know. Not all students need to be tested every year, as long as teachers are deeply involved in assessing student progress. Instead of practicing test-taking and taking lots of standardized tests, students could be playing, reading, writing, doing original research, and working on projects.

Excessive testing doesn’t cause obesity directly, but it has led schools to reduce or eliminate recess, which has in turn contributed to the rise of weight problems. Unfortunately, poverty correlates with excessive testing, weight problems, and reduced recess, meaning that poor children once again draw the short straw.

If we hold schools accountable for both academic results and hours of free play, educators will be forced to cut back on testing and test-prep drilling. That simple change—call it “Measuring What Matters”–is a sensible education policy that should also produce measurable health benefits. And since kids who get to exercise regularly tend to do better academically, we should also see improvements there as well.

Bottom line: This is totally on us!  If we truly want healthier children, we have to cut back on testing and test-prep and bring back recess and free play.